The distribution of ciliates in a Mexican coastal lagoon was studied. The 4 goals were to: examine small-scale (<100 m) patches; indicate how geostatistical techniques can be used to examine these patches; make inferences concerning ciliate distribution and behaviour in the lagoon using geostatistical techniques; and assess geostatistics as a method for modelling ciliate distributions. Underlying these goals we attempt to make geostatistical techniques accessible to the non-expert. We provide an overview of the methodology, references to the geostatistical literature, and use our system as an example. Ciliates were sampled in a 40 × 40 m grid, divided at 10 m intervals; the grid was further divided into subsets, to determine 1 to 10 m scale variation. Between 30 and 35 points were sampled on 2 occasions (January and October). Ciliates were preserved with Lugol's iodine; abundance and species composition were determined by standard inverted microscopy. The work focused on 4 abundant ciliate species. We indicate, using the variographic analysis, that the abundance of 3 of the 4 ciliates is neither randomly nor homogeneously distributed, but exhibits a structured small-scale patchy distribution. We indicate that species-specific patterns of patchiness exist in stratified and in mixed waters, supporting the notion of behavioural niche-separation of planktonic ciliates. Patches of <13, <18, and < 77 m were formed by Lohmaniella oviformis, Tintinnopsis sp. and Strombidium sp., respectively. In contrast, Pleuronema sp. formed patches below the detection limits of the analysis (<1 m). Using geostatistical techniques, we established variograms and used them to model ciliate distribution and predict ciliate behaviour. Distribution maps were then generated that depicted the shape, distinctness, and gradient of the different patches. After analysing the data, we proposed a working definition of a 'ciliate patch': regions with abundance above the cut-off of the upper quartile from the kriging prediction model. Finally, error-maps were developed, indicating the coefficient of variation of the predicted distributions. We conclude that geostatistical analysis is a powerful tool to examine microzooplankton at small-scales, and we support its further application in the field.

Myrionecta rubra, a ubiquitous planktonic ciliate, has received much attention due to its wide distribution, occurrence as a red tide organism, and unusual cryptophyte endosymbiont. Although well studied in coastal waters, M. rubra is poorly examined in the open ocean. In the Irminger Basin, North Atlantic, the abundance of M. rubra was 0-5 cells/ml, which is low compared with that found in coastal areas. Distinct patchiness (100 km) was revealed by geostatistical analysis. Multiple regression indicated there was little relationship between M. rubra abundance and a number of environmental factors, with the exception of temperature and phytoplankton biomass, which influenced abundance in the spring. We also improve on studies that indicate distinct size classes of M. rubra; we statistically recognise four significantly distinct width classes (5-16, 12-23, 18-27, 21-33 microm), which decrease in abundance with increasing size. A multinomial logistic regression revealed the main variable correlated with this size distribution was ambient nitrate concentration. Finally, we propose a hypothesis for the distribution of sizes, involving nutrients, feeding, and dividing of the endosymbiont.

Myrionecta rubra patchiness in a Mexican coastal lagoon was studied. The 3 objectives were to (1) characterize the spatial distribution of M. rubra patches through time; (2) characterize and model the spatial distribution of M. rubra at scales ranging from m to km, and from 1 wk to more than 1 yr; and (3) to place the patchiness patterns of M. rubra into an ecological context. Geostatistical analysis was applied to data collected from simple and nested sampling grids in different seasons; autocorrelation analysis was used to detect temporal regularities over 55 wk. Classical statistics were applied to data from 10 sites in the lagoon to identify trends relating ciliate abundance to environmental conditions. Patches were detected and characterized using empirical variograms and modelled by omnidirectional Gaussian and exponential functions. For most of the analysis variance was low in the nugget parameter, indicating a strong spatial resolution of the data, and the range parameter indicated that M. rubra formed patches of 10, 20, 80, 130, and 170 m. Spatial analysis using hierarchical grids produced a more detailed assessment of patches than single grids alone. Conditional simulation of patches indicated the presence of a > 2 km patch covering most of the western lagoon. Patch densities varied from between 4 and 700 cells ml -1 . M. rubra abundance exhibited a temporal, pulse-like pattern; autocorrelation revealed a 13 wk periodicity. At the lagoonal level, multiple regression revealed a trend towards higher abundance in the north-west of the lagoon and a decrease during the dry season. Finally, we speculate on the forces causing heterogeneity at large (>1000 m), meso (100 to 1000 m), and fine (1 to 100 m) scales by examining physical-chemical environmental factors and physiological behavioural properties of the ciliate and its potential predators. We propose that M. rubra patches originate by fragmentation of larger patches, growth of smaller patches, and physical-behavioural aggregation of cells.

This study focuses on a large region of the open ocean where we predict that microzooplankton significantly influence foodweb structure over much of the year. The Irminger Sea exhibits low primary production that is generally poor for copepod production; in such waters, ciliates and other microzooplankton are major grazers of primary production and contribute significantly to the diets of holo-and mero-mesozooplankton. Surface plankton samples were collected during an extensive survey across the basin and along one transect at several depths, over 3 seasons (winter, spring, summer), but not including the spring bloom. Microzooplankton and phytoplankton samples were fixed with Lugol's solution and microscopically enumerated for species abundance; biomass was determined from cell volumes. Basin-scale distributions of abundance, biomass, and production were examined by geostatistical and multidimensional scaling methods. Dominance of the <10 µm phytoplankton suggests that this should be a microzooplankton-dominated food web. Ciliates and heterotrophic dinoflagellates are abundant in terms of numbers and biomass; heterotrophic dinoflagellates are more abundant than ciliates, but are less dominant in terms of biomass. Using ciliates as a proxy for all microzooplankton, we suggest that there are seasonal patterns in occurrence, and there is no basin-scale patchiness related to hydrographic features. We suggest that ciliate production is sufficient to account for the removal of 15 to 30% of the <10 µm primary production. If heterotrophic dinoflagellates were included in these estimates, removal may be doubled (i.e. 30 to 60%). We therefore contend that microzooplankton are major phytoplankton consumers in the system and should be carefully parameterised in models of this region.

Recognising that species interact across a range of spatial scales, we explore how landscape structure interacts with temperature to influence persistence. Specifically, we recognise that few studies indicate thermal shifts as the proximal cause of species extinctions; rather, species interactions exacerbated by temperature result in extinctions. Using microcosm‐based experiments, as models of larger landscape processes, we test hypotheses that would be problematic to address through field work. A text‐book predator–prey system (the ciliates Didinium and Paramecium) was used to compare three landscapes: an unfragmented landscape subjected to uniform temperatures (10, 20, 30°C); a fragmented landscape (potentially hosting metapopulations) subjected to these three temperatures; and a fragmented landscape subjected to a spatial temperature gradient (∼ 10 to 30°C) – despite the prevalence of natural temperature ecoclines this is the first time such an analysis has been conducted. Initial thermal response‐analysis (growth, mortality, and movement measured between 10 and 30°C) suggested that as temperature increased, the predator might drive the prey to extinction. Thermal preferences (measured at 5 temperatures between 10 and 30°C), indicated that both predator and prey preferred warmer temperatures, with the predator exhibiting the stronger preference, suggesting that cooler regions might act as a prey‐refuge. The landscape level observations, however, did not entirely support the predictions. First, in the unfragmented landscape, increased temperature led to extinctions, but at the highest temperature (where the predator growth can be reduced) the prey survived. Second, at high temperatures the fragmented landscape failed to host metapopulations that would allow predator–prey persistence. Third, the thermal ecocline did not provide heterogeneity that improved stability; rather it forced both species to occupy a smaller realized space, leading toward extinctions. These findings reveal that temperature‐impacted rates and temperature preferences combine to drive predator–prey dynamics and persistence across landscapes.