Principles and methods for automated palynology

Pollen grains are microscopic so their identification and quantification has, for decades, depended upon human observers using light microscopes: a labour-intensive approach. Modern improvements in computing and imaging hardware and software now bring automation of pollen analyses within reach. In this paper, we provide the first review in over 15 yr of progress towards automation of the part of palynology concerned with counting and classifying pollen, bringing together literature published from a wide spectrum of sources. We
consider which attempts offer the most potential for an automated palynology system for universal application across all fields of research concerned with pollen classification and counting. We discuss what is required to make the datasets of these automated systems as acceptable as those produced by human palynologists, and present suggestions for how automation will generate novel approaches to counting and classifying pollen that have hitherto been unthinkable.

Pollen image classification using the classifynder system: Algorithm comparison and a case study on New Zealand honey

We describe an investigation into how Massey University’s 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 University’s 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 Classifynder’s 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.

A comparison of classification algorithms within the Classifynder pollen imaging system

We describe an investigation into how Massey University's 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 University's pollen reference collection (2890 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. In addition to the Classifynder's 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. © 2013 AIP Publishing LLC.

Differentiation between Neotropical rainforest, dry forest and savannah ecosystems by their modern pollen spectra and implications for the fossil pollen record

Accurate differentiation between tropical forest and savannah ecosystems in the fossil pollen record is hampered by the combination of: i) poor taxonomic resolution in pollen identification, and ii) the high species diversity of many lowland tropical families, i.e. with many different growth forms living in numerous environmental settings. These barriers to interpreting the fossil record hinder our understanding of the past distributions of different Neotropical ecosystems and consequently cloud our knowledge of past climatic, biodiversity and carbon storage patterns. Modern pollen studies facilitate an improved understanding of how ecosystems are represented by the pollen their plants produce and therefore aid interpretation of fossil pollen records. To understand how to differentiate ecosystems palynologically, it is essential that a consistent sampling method is used across ecosystems. However, to date, modern pollen studies from tropical South America have employed a variety of methodologies (e.g. pollen traps, moss polsters, soil samples). In this paper, we present the first modern pollen study from the Neotropics to examine the modern pollen rain from moist evergreen tropical forest (METF), semi-deciduous dry tropical forest (SDTF) and wooded savannah (cerradão) using a consistent sampling methodology (pollen traps). Pollen rain was sampled annually in September for the years 1999–2001 from within permanent vegetation study plots in, or near, the Noel Kempff Mercado National Park (NKMNP), Bolivia. Comparison of the modern pollen rain within these plots with detailed floristic inventories allowed estimates of the relative pollen productivity and dispersal for individual taxa to be made (% pollen/% vegetation or ‘p/v’). The applicability of these data to interpreting fossil records from lake sediments was then explored by comparison with pollen assemblages obtained from five lake surface samples.

Pollen and diatom analysis on varved sediments in Lake Jues (Harz Mountains, Germany)

Studies combining sedimentological and biological evidence to reconstruct Holocene climate beyond the major changes, and especially seasonality, are rare in Europe, and are nearly completely absent in Germany. The present study tries to reconstruct changes of seasonality from evidence of annual algal successions within the framework of well-established pollen zonation and 14C-AMS dates from terrestrial plants. Laminated Holocene sediments in Lake Jues (10°20.70' E, 51°39.30' N, 241 m a.s.l.), located at the SW margin of the Harz Mountains, central Germany, were studied for sediment characteristics, pollen, diatoms and coccal green algae. An age model is based on 21 calibrated AMS radiocarbon dates from terrestrial plants. The sedimentary record covers the entire Holocene period. Trophic status and circulation/stagnation patterns of the lake were inferred from algal assemblages, the subannual structure of varves and the physico-chemical properties of the sediment. During the Holocene, mixing conditions alternated between di-, oligo- and meromictic depending on length and variability of spring and fall periods, and the stability of winter and summer weather. The trophic state was controlled by nutrient input, circulation patterns and the temperature-dependent rates of organic production and mineralization. Climate shifts, mainly in phase with those recorded from other European regions, are inferred from changing limnological conditions and terrestrial vegetation. Significant changes occurred at 11,600 cal. yr. BP (Preboreal warming), between 10,600 and 10,100 cal. yr. BP (Boreal cooling), and between 8,400 and 4,550 cal. yr. BP (warm and dry interval of the Atlantic). Since 4,550 cal. yr. BP the climate became gradually cooler, wetter and more oceanic. This trend was interrupted by warmer and dryer phases between 3,440 and 2,850 cal. yr. BP and, likely, between 2,500 and 2,250 cal. yr. BP.

Pollen and macrofossils profiles and age determination from the Pittsburg Basin in the lowlands of south-central Illinois

Pollen and macrofossil evidence for the nature of the vegetation during glacial and interglacial periods in the regions south of the Wisconsinan ice margin is still very scarce. Modern opinions concerning these problems are therefore predominantly derived from geological evidence only or are extrapolated from pollen studies of late Wisconsinan deposits. Now for the first time pollen and macrofossil analyses are available from south-central Illinois covering the Holocene, the entire Wisconsinan, and most probably also Sangamonian and late Illinoian time. The cores studied came from three lakes, which originated as kettle holes in glacial drift of Illinoian age near Vandalia, Fayette County. The Wisconsinan ice sheet approached the sites from the the north to within about 60 km distance only. One of the profiles (Pittsburg Basin) probably reaches back to the late Illinoian (zone 1), which was characterized by forests with much Picea. Zone 2, most likely of Sangamonian age, represents a period of species-rich deciduous forests, which must have been similar to the ones that thrive today south and southeast of the prairie peninsula. During the entire Wisconsinan (14C dates ranging from 38,000 to 21,000 BP) thermophilous deciduous trees like Quercus, Carya, and Ulmus occurred in the region, although temporarily accompanied by tree genera with a more northerly modern distribution, such as Picea, which entered and then left south-central Illinois during the Woodfordian. Thus it is evident that arctic climatic conditions did not prevail in the lowlands of south-central Illinois (about 38°30' lat) during the Wisconsinan, even at the time of the maximum glaciation, the Woodfordian. The Wisconsinan was, however, not a period of continuous forest. The pollen assemblages of zone 3 (Altonian) indicate prairie with stands of trees, and in zone 4 the relatively abundant Artemisia pollen indicates the existence of open vegetation and stands of deciduous trees, Picea, and Pinus. True tundra may have existed north of the sites, but if so its pollen rain apparently is marked by pollen from nearby stands of trees. After the disappearance of Pinus and Picea at about 14,000 BP (estimated!), there developed a mosaic of prairies and stands of Quercus, Carya, and other deciduous tree genera (zone 5). This type of vegetation persisted until it was destroyed by cultivation during the 19th and 20th century. Major vegetational changes are not indicated in the pollen diagram for the late Wisconsinan and the Holocene. The dating of zones 1 and 2 is problematical because the sediments are beyond the14C range and because of the lack of stratigraphic evidence. The zones dated as Illinoian and Sangamonian could also represent just a Wisconsinan stadial and interstadial. This possibility, however, seems to be contradicted by the late glacial and interglacial character of the forest vegetation of that time.

Pollen profiles from a Late Palaeolithic site in the community of Bad Buchau-Kappel, Baden-Wuerttemberg, Germany

Excavations were carried out in a Late Palaeolithic site in the community of Bad Buchau-Kappel between 2003 and 2007. Archaeological investigations covered a total of more than 200 m**2. This site is the product of what likely were multiple occupations that occurred during the Late Glacial on the Federsee shore in this location. The site is situated on a mineral ridge that projected into the former Late Glacial lake Federsee. This beach ridge consists of deposits of fine to coarse gravel and sand and was surrounded by open water, except for a connection to the solid shore on the south. A lagoon lay between the hook-shaped ridge and the shore of the Federsee. This exposed location provided optimal access to the water of the lake. In addition, the small lagoon may have served as a natural harbor for landing boats or canoes. Sedimentological and palynological investigations document the dynamic history of the location between 14,500 and 11,600 years before present (cal BP). Evidence of the deposition of sands, gravels and muds since the Bølling Interstadial is provided by stratigraphic and palynological analyses. The major occupation occurred in the second half of the Younger Dryas period. Most of the finds were located on or in the sediments of the ridge; fewer finds occurred in the surrounding mud, which was also deposited during the Younger Dryas. Direct dates on some bone fragments, however, demonstrate that intermittent sporadic occupations also took place during the two millennia of the Meiendorf, Bølling, and Allerød Interstadials. These bones were reworked during the Younger Dryas and redeposited in the mud. A 14C date from one bone of 11,600 years ago (cal BP) places the Late Palaeolithic occupation of the ridge at the very end of the Younger Dryas, which is in agreement with stratigraphic observations. Stone artifacts, numbering 3,281, comprise the majority of finds from the site. These include typical artifacts of the Late Palaeolithic, such as backed points, short scrapers, and small burins. There are no bipointes or Malaurie-Points, which is in accord with the absolute date of the occupation. A majority of the artifacts are made from a brown chert that is obtainable a few kilometers north of the site in sediments of the Graupensandrinne. Other raw materials include red and green radiolarite that occur in the fluvioglacial gravels of Oberschwaben, as well as quartzite and lydite. The only non-local material present is a few artifacts of tabular chert from the region near Kelheim in Bavaria. A unique find consists of two fragments of a double-barbed harpoon made of red deer antler, which was found in the Younger Dryas mud. It is likely, but not certain, that this find belongs to the same assemblage as the numerous stone artifacts. Although not numerous, animal bones were also found in the excavations. Most of them lay in sediments of the Younger Dryas, but several 14C dates place some of these bones in earlier periods, including the Meiendorf, Bølling, and Allerød Interstadials. These bones were reworked by water and redeposited in mud sediments during the Younger Dryas. As a result, it is difficult to attribute individual bones to particular chronological positions without exact dates. Species that could be identified include wild horse (Equus spec.), moose or elk (Alces alces), red deer (Cervus elaphus), roe deer (Capreolus capreolus), aurochs or bison (Bos spec.), wild boar (Sus scrofa), as well as birds and fish, including pike (Esox Lucius).

Quaternary pollen analysis: recent progress in palaeoecology and palaeoclimatology

During the last decade Quaternary pollen analysis has developed towards improved pollen-taxonomical precision, automated pollen identification and more rigorous definition of pollen assemblage zones. There have been significant efforts to model the spatial representation of pollen records in lake sediments which is important for more precise interpretation of the pollen records in terms of past vegetation patterns. We review the difficulties in matching modelled post-glacial plant migration patterns with pollen-based palaeorecords and discuss the potential of DNA analysis of pollen to investigate the ancestry and past migration pathways of the plants. In population ecology there has been an acceleration of the widely advocated conceptual advance of pollen-analytical research from vaguely defined ‘environmental reconstructions’ towards investigating more precisely defined ecological problems aligned with the current ecological theories. Examples of such research have included an increasing number of investigations about the ecological impacts of past disturbances, often integrating pollen records with other palaeoecological data. Such an approach has also been applied to incorporate a time perspective to the questions of ecosystem restoration, nature conservation and forest management. New lines of research are the use of pollen analysis to study long-term patterns of vegetation diversity, such as the role of glacial-age vegetation fragmentation as a cause of Amazonian rain forest diversity, and to investigate links between pollen richness and past plant diversity. Palaeoclimatological use of pollen records has become more quantitative and has included more precise and rigorous testing of pollen-climate calibration models with modern climate data. These tests show the approximate nature of the models and warn against a too straightforward climatic interpretation of the small-scale variation in reconstructions. Pollenbased climate reconstructions over the Late Glacial–early Holocene boundary have indicated that pollen-stratigraphical changes have been rapid with no evidence for response lags. This does not rule out the possibility of migrational disequilibrium, however, as the rapid changes may be mostly due to nonmigrational responses of existing vegetation. It is therefore difficult to assess whether the amplitude of reconstructed climate change reflects real climate change. Other outstanding problems remain the obscure relationship of pollen production and climate, the role of human impact and other nonclimatic factors, and nonanalogue situations.

Pollen morphology in Brazilian species of Codonanthe (Mart.) Hanst. and Nematanthus Schrader (Gesneriaceae)

Pollen morphology of 24 Brazilian species of Codonanthe and Nematanthus (Gesneriaceae) was investigated by means of light microscopy, scanning and transmission electron microscopy, in search of new characters that might contribute to taxonomic characterisation of the genera. The pollen grains are tricolpate or tricolporate, semitectate, reticulate, with simplicolumellate muri, sometimes with microreticulum and perforations, or duplicolumellate muri ring-shaped around lumina. The species studied were grouped into five pollen types on the basis of aperture characteristics and details of surface ornamentation. Sub-types were defined with reference to presence or absence of microreticulum and perforations around the lumina. Based on analysis of the morphology and measurements of pollen grains, we verified a close relationship among the genera, but can nevertheless discriminate their species. © 2013 Copyright Collegium Palynologicum Scandinavicum.