Diatom Gliding Motility. Группа авторов
with their short raphe system, typically described as slightly (< 2 µm/sec [8.31]) or weakly motile (2 to 4 µm/sec [8.31]) (see eunotioid taxa in DONA [8.21] – Furey [8.26]), contrasts with that of more motile forms like naviculoid, nitzschioid, or surirelioid diatoms with more extensive raphe systems, typically described as moderately to highly motile. Examination of motility in Eunotia species may provide unique insight into motility in diatoms, especially for diatoms with more complex raphe systems. A search for the terms “motile” or “move” in florae focused on Eunotia (e.g., [8.20] [8.28] [8.48] [8.51]) and in >40 manuscripts with descriptions of Eunotia taxa new to science revealed little to no mention of motility. This chapter focuses on motility in the diatom genus Eunotia, but does not cover cellular or biomechanical details around the mechanisms of movement, as other chapters in this book discuss these aspects at length.Figures 8.23–8.27 Schematic representation of some of the movement types for Eunotia. (Figure 8.23) Schematic of forward movement – apical displacement where cells tilt slightly so the anterior ends of the valves remain in contact with the surface and the posterior ends become slightly raised (**schematic modeled after Palmer [8.60] plate vi. fig. 2, and Bertrand [8.6] fig. 1**). (Figure 8.24) Valve in girdle view. (Each raphe branch labeled after Bertrand [8.6] (**see also Harbich [8.30]**). Black arrow following the line of the raphes on B to C apices represents diagonal line of direction, where the raphe on the C end becomes active. (Figure 8.25) Black horizontal arrow represents diagonal line of direction. Bidirectional arrow shows transition between raphes involved in forward motion (**schematic modeled after Bertrand [8.6], fig. 1**). (Figure 8.26) Schematic of a vertical polar pivot which can return a cell in girdle view, dorsal-side down (a) to ventral-side down (b,c) where a cell can then continue forward movement (c) (**schematic modeled after Palmer [8.60], plate vi. fig. 2, and Bertrand ([8.6], fig. 5]**). (Figure 8.27) Schematic of a horizontal, polar pivot (a,b) to show direction of raphe activity (straight arrows, A and B). Note the cell is depicted ventral side up so the raphe branches are visible (rather than dorsal side up as the movement occurs). Curved arrows show direction of rotation. Dot represents the pivot point. (**Modified after images from Harbich [8.30]). See additional schematics in Bertand [8.5].
9 Chapter 9Figure 9.1 Gliding cells of Nitzschia sigmoidea with stalked epiphytes of Pseudostaurosira parasitica: a single epiphyte in connective view (a), in valve view (b), and two epiphytes attached to the same frustule (c), which can be seen in movement [9.39].Figure 9.2 Cells of Nitzschia sigmoidea with adnate epiphytes of Fallacia helensis (OM): single epiphyte in connective view and host in valve view (a), four epiphytes in valve view and host in connective view (b), which can be seen in movement [9.43].Figure 9.3 Cells of Nitzschia sigmoidea with many epiphytes (OM): epiphytes of Amphora copulata on a still gliding host (a), epiphytes of Amphora copulata on a dividing host (b), epiphytes of Amphora copulata and Pseudostaurosira parasitica on the same host (c), which can be seen in movement [9.41].Figure 9.4 Focus on epiphytes of Nitzschia sigmoidea (SEM). Adnate, Fallacia helensis, and stalked, Pseudostaurosira parasitica, epiphytes on the same host (a), two cells of Pseudostaurosira parasitica still associated after division (b), two superimposed cells of Fallacia helensis after division and a third single one attached on the edge of the frustule (c), apex of a cell of Pseudostaurosira parasitica with the mucilaginous pad secreted for adhesion (d), individual of Amphora copulata (e) and internal view of one valve of Amphora sp., probably Amphora copulata var. epiphytica Round & Kyung Lee, considering the almost circular areolae on the ventral side (f). Scale bars indicate 10 µm, except in 9.4d (=5 µm).Figure 9.5 Variations in the specific composition of epiphytes on Nitzschia sigmoidea between two sampling sites located on two connected rivers (up- and downstream sites), expressed as the occurrence of three epidiatomic species on frustules of N. sigmoidea. In fact, through the observation of fresh material, N. sigmoidea could not be strictly distinguished from the close species N. vermicularis. Both species were present in each site (see [9.44] for details).Figure 9.6 Sigmoid frustules of Nitzschia sigmoidea (NSIO) and Gyrosigma attenuatum (GYAT) (OM, H2O2 treated material). Two species co-occurring in rivers samples with similar abundance, valve length and motility. However, Gyrosigma attenuatum was never seen with epiphytes.
10 Chapter 10Figure 10.1 Drawing adapted from O.F. Müller (1783, translated in [10.54]), who was the first to characterize Bacillaria colonies. Examples 1 through 8 show the various states of expansion and contraction (dynamic phenotypes) of colonies.Figure 10.2 Bacillaria close-up images of single cells using scanning electron microscopy (SEM). (a) a whole valve seen from the inside. (b) close up of the same, middle section. The horizontal slit is the raphe. It lacks a central node. (c) Tip of the inside of a valve. (d) Middle section of a valve, exterior view. Note the raphe is a slit through the whole valve. (e): External view of the tip of a valve [10.33]. (Reprinted with permission of Amgueddfa Cymru, National Museum Wales.)Figure 10.3 Demonstration of the image tracking procedure. (a) definition of tracked feature (white ellipse within a cell). (b) labeled (numbered) cells with relative measurements provided in red. The determined coordinates refer to a target in the middle of the template. The target can be moved and placed on the apex of the diatom being tracked. Then the coordinates of the apex are captured. This position is indicated in Figure 10.3a by a mark and a vertical line. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Scale bars 50 µm.Figure 10.4 A diagram showing the five points on a sample cell (two ends, midpoint of the transverse line, and edges of the cell).Figure 10.5 Point A is on the edge (vertical direction). The gradient direction is normal to the edge. Points B and C are in gradient directions. So, point A is checked with point B and C to see if it forms a local maximum. If so, it is considered for the next stage, otherwise, it is suppressed (set to zero). The result is a binary image with “thin edges.”Figure 10.6 Diagram showing an example of hysteresis thresholding and the labeled edge relative to the “sure edge” threshold (Vmax).Figure 10.7 An example of feature identification performance for the Watershed Segmentation algorithm (left, red boundary) and Canny Edge Detection algorithm (right, white boundary). Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Scale bars 50 µm.Figure 10.8 An example of feature identification training (purple rectangles) for the deep learning approach on a single set of cells. Notice the resolution of the colony. An example of correct performance is shown in Figure 10.5. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Scale bar 50 µm.Figure 10.9 Rank-order analysis of bounding box (cell) sizes (area) across the dataset. The area is measured in pixels squared. Image scale: 38.36 μm per cm, or 0.325 μm per pixel.Figure 10.10 Rank-order analysis of height (blue) and width (red) of bounding boxes (cells) across the dataset. The area is measured in pixels squared. Image scale: 38.36 μm per cm, or 0.325 μm per pixel.Figure 10.11 (Top) location of centroids in normalized coordinate space in selected dataset for static analysis. (Bottom) First two principal components from PCA analysis (PC1 represents horizontal position, while PC2 represents vertical position) of coordinates representing the x,y position for all four edges of each bounding box using the selected datasets for static analysis. Image scale: 38.36 μm per cm, or 0.325 μm per pixel.Figure 10.12 An example of feature identification optimization procedures implemented in DeepLabv3. GRAY: no optimization applied, RED: Optimization #1, BLUE: Optimization #2. Given an initial number of training frames (y-axis), the non-optimized procedure (originally detected) will yield a certain number of boxes (x-axis). Applying various optimization procedures generally leads to a decreased number of boxes per frame for both low and high numbers of boxes.Figure 10.13 Four examples of how the identified features map to two different images (a, c, e, g) of a Bacillaria colony. Points (b, d, f, h) represent the centroids for all bounding boxes identified in images a, c, e, and g, respectively. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Scale bars 50 µm.Figure 10.14 Three examples of how images of a Bacillaria colony are converted into a skeleton image. (Top Row) light microscopy images, (Middle Row) thin skeletonization based on a procedure implemented in GIMP, (Bottom Row) thick skeleton based on a procedure implemented in GIMP. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Image scale: 38.36 μm per cm, or 0.325 μm per pixel. Scale bars 50 µm.Figure 10.15 Examples of relative movement of cells in a sample colony. (a) Comparisons between changes of position for cell #2 relative to cell #1 (red) and cell #3 relative to cell #2 (blue). (b) Comparisons between changes of position