RU2541462C1 - Method for identifying living and dead mesozooplankton in seawater samples - Google Patents

Abstract

FIELD: biotechnology.

SUBSTANCE: invention refers to a method for identifying living and dead mesozooplankton in seawater samples, which involves taking samples, staining the organisms with suitable colouring material, giving a visual estimation of the colour intensity of the units under the microscope, which is combined with microphotographying the units with an adjustable camera without changing the settings keeping throughout a photographic session of at least one sample; thereafter colour and brightness specifications average for each unit are measured in the formed images with using a painting program, e.g. Adobe Photoshop package, and the units are referred to living or dead by a discriminative analysis of the varied digital values.

EFFECT: improving the method.

Description

The invention relates to hydrobiology and can be used for differentially accounting for living and dead organisms in marine samples.

The quantitative ratio of living and dead organisms in samples of mesozooplankton is one of the basic indicators of community mortality, which can be used to study the patterns of functioning of marine ecosystems, their bioproductivity and the circulation of substances, as well as to assess the ecological state of water areas.

Two main methodological approaches are known for identifying living and dead zooplankton organisms: visual assessment and staining. Visual analysis evaluates the external and internal state of the body on a fixed material, in other words, reveals the presence or absence of signs of decomposition. This approach began to be applied in the 30s of the XX century. and continue to be used at present [1,2], however, it is not without serious flaws: firstly, the researcher needs a lot of experience with zooplankton; secondly, - reliable detection of signs of decomposition is possible only 2-3 hours after the death of the body (in some cases and later), which inevitably leads to underestimation of the proportion of dead individuals in the sample.

A known method, which was developed in the 70s of the last century, is based on staining plankton with various dyes [3]. The main drawback of this approach is that the degree of staining of organisms in natural samples can vary greatly. Consequently, the researcher’s conclusion that the organism with a “doubtful” color belongs to one or another category is subjective. If the number of such organisms in the sample is large, reliable and reproducible results cannot be obtained.

The basis of the invention, “A method for the identification of living and dead organisms of mesozooplankton in marine samples,” is tasked with obtaining reliable, objective and reproducible results by classifying organisms by degree of color.

The problem is solved in that in the method, by performing a visual assessment under a microscope of an individual’s color intensity, they are simultaneously microphotographed with the camera settings in manual mode, keeping these settings unchanged during photography of at least one sample, after which they are performed for the obtained images measuring average color and brightness characteristics for each individual using a raster graphics editor, such as the Adobe Photoshop software package, and classify the obtained values using skriminantny analysis.

Between the listed essential features and the expected technical effect, the following causal relationship is present. Regardless of which dye is used to label living / dead organisms, the presence or absence of color and / or brightness is fundamental. The color (tonal) and brightness characteristics of each of the analyzed individuals, presented in digital form, make it possible to use discriminant analysis to objectively and reliably divide the entire studied population into two classes - living and dead individuals.

The invention is illustrated by illustrations. In FIG. Figure 1 presents Digital images of copepods stained with fluorescein diacetate - FDA (A, B) and neutral red - NK (C, Γ) · A and C - live individuals, B and D - dead individuals. The dotted line marks the region of the most intense color - cephalothorax; in FIG. 2 - Measurement of the averaged color characteristics of a copepod specimen stained with NK using the color palette of Adobe Photoshop CS2; in FIG. 3 - Identification of clusters of living and dead organisms after their staining with FDA (A) and NK (B). The data were obtained for the culture of copepods (Calanipeda aquae dulcís) alive and killed by temperature shock. In FIG. 4 - Data format in the Statistica 5.0 package; in FIG. 5 - The main results of the analysis; in FIG. 6 - Clusters of points corresponding to the classes of organisms L (living, □), D (dead, ∇) and Q (doubtful, +), on 2-parameter diagrams with different independent variables: A - green (Green) and blue (Blue) ; B - green and shade (Hue); In - green and red (Red). Diagram B presents the results of the classification of individuals of class Q using discriminant analysis: ■ - classified as living; ▼ - assigned to the dead. The arrow indicates the only “outlier” of class D; the dashed line is the conditional boundary between the classes L and D; in FIG. 7 - Fragment of the table of posterior probabilities of individuals belonging to a particular class.

The method is implemented as follows:

1. Mesozooplankton sampling and its color for identification of living / dead organisms are carried out in accordance with the methods provided for these purposes.

2. The study of a representative sample of organisms under a microscope:

a) the assignment of each of the individuals to one of the three classes according to visual signs (color intensity):

- “Living” (L);

- “Dead” (D);

- “Doubtful” (Q).

Visual assessment of the color intensity of individuals is made in the field of view of the microscope in accordance with the method of using the dye. Each individual belongs to one of two classes - “Living” (L) or “Dead” (D). If this is difficult or impossible, the individual is classified as “doubtful” (Q).

b) obtaining digital images of each individual using microphotography.

Microphotography of stained organisms is carried out by a digital camera in color. Automatic camera mode (automatic exposure selection) is not recommended because if only bright (for example, living organisms painted with FDA) or only dark objects (for example, dead organisms close in brightness to a dark background fall into the field of view), then the auto-correction of images produced by the camera will distort color and brightness (contrast of the object with the background ) characteristics of objects, and, as a consequence of this, comparison of such images will be impossible. It is recommended to select the camera settings in manual mode in such a way as to avoid, firstly, overexposure on the brightest objects (i.e. loss of information about brightness and color) and, secondly, merging objects with the background (for example, unpainted objects in the dark field of a luminescent microscope), and keep these settings unchanged during photography of at least one sample. This is a guarantee that images of both living and dead organisms of the same sample are obtained under the same conditions and, therefore, can be used for further analysis. The less often you have to change the camera settings, adapting them to each specific dye, species of organisms or a series of samples, the higher the reliability of the results. At the same time, the individual characteristics of cameras used by different researchers (differences in noise level, offset in white balance, etc.) cannot affect the quality of the results. This means that the analysis of the same zooplankton sample, carried out on different equipment, but in accordance with the proposed method, will give close results of assessing the ratio of living and dead organisms in the sample.

3. Measurement of average for each individual color and brightness characteristics (color models HSB, RGB, CMYK, LAB) in a graphical editor.

As a software for further work with images, any raster graphics editor is used, which allows you to determine the basic color and brightness characteristics of any of the image pixels in accordance with three color models: HSB (H - color tone, S - saturation, B - brightness), RGB (R - red, G - green, B - blue) and CMYK (C - cyan, M - magenta, Y - yellow, K - black). For these purposes, you can use, for example, the Adobe Photoshop software package or a number of free graphic editors that are freely distributed on the Internet, for example, Eyedropper 4.0 beta, Pixel Pick 1.5, SI ColorPicker 1.0, etc.

For each analyzed individual, one set of HSB characteristic values is obtained; therefore, localization and the method of color sampling within the boundaries of the image of the organism are important. Selection should be made in areas with the highest color intensity (for example, for copepods stained with FDA or neutral red, this is cephalothorax; highlighted in Fig. 1, A and B). In unpainted organisms, the same zone is used (Fig. 1, B and D). If the image of the region of interest to the researcher is heterogeneous in color and brightness, has spots and / or granularity (as, for example, this is often the case with copepods, Fig. 1, B), it is recommended that you first select the entire colored region (Lasso Tool, Lasso Tool, in Adobe Photoshop) and average its luminance and color characteristics, and then measure them.

4. The data obtained are tabulated in a format suitable for discriminant analysis.

Upon completion of measurements of color characteristics of all individuals of the sample (see Fig. 2), the obtained digital data is summarized in a table in which each row corresponds to one individual, the number of lines is equal to the number of individuals. Each individual assigned to one of the classes I, D or Q (CLASS categorical variable) is characterized by color variables, the maximum number of which can be 13 (RGB, HSB, CMYK, LAB). The best way is to limit yourself to 6 variables of the RGB and HSB models. To work with each specific dye, as a rule, a small set of variables is required. For example, when staining zooplankton with an FDA dye having green fluorescence, it is sufficient to measure saturation and brightness (respectively, S and B from HSB). It is possible to reduce the dimension of the data (i.e., the number of analyzed variables) after it becomes clear which variables allow the most efficient detection of the classes of living and dead organisms. On 2-parameter diagrams, the latter will look like well distinguishable clusters of points (Fig. 3, A). If the clusters L and D cannot be detected in any of the diagrams and, as a result of this, the choice of variables that are large for the classification of organisms is difficult, the data dimension is reduced using discriminant analysis (see below).

5. The use of discriminant analysis for:

a) reducing the dimensionality of data (step-by-step analysis with the inclusion of variables);

b) constructing a classification of objects using classes L and D as a training sample;

c) assignment of each of the organisms of class Q to classes L or D in accordance with the discriminant model.

Classification of organisms with a weakly pronounced coloration. In natural samples, a large number of organisms are usually present, the color intensity of which is small (class Q), and which are difficult to attribute to the living or the dead using visual analysis. In the diagrams, they form a point cloud, which is superimposed on the L and D clusters (see the example below), which complicates the choice of color variables that would provide an objective and reliable classification. To solve both problems, we propose using discriminant analysis, an effective method for constructing a classification using a training sample. The latter should include organisms that were easily assigned by the researcher to classes L or D during microscopy of the sample. In the data pivot table, information about the body belonging to a particular class is contained in the CLASS categorical variable. In discriminant analysis, it is used as a grouping dependent variable, and color characteristics (Hue, Saturation, Brightness, Red, Green, etc.), measured in the metric scale, are considered independent variables. Reducing the dimensionality of data can be done in a step-by-step analysis of discriminant functions. At each step, all independent variables are scanned, and one of them is found that makes the greatest contribution to the difference between classes. The discriminant analysis model, which is built on the basis of the training sample, allows us to establish with certain certainty that the organisms with the controversial coloration (class Q) belong to classes L or D. An example of the application of discriminant analysis to solving this kind of problem is given in the Example.

Example.

Determination of the proportion of living copepods in marine zooplankton after staining with neutral red (NK).

Samples of marine zooplankton were taken in the coastal waters of the city of Sevastopol (Black Sea) and stained with NK in accordance with standard methods. Living organisms stained with NK acquired a red color, while the dead remained pale yellow. Organisms were photographed using a Nikon Eclipse TS100-F microscope equipped with an Ikegami ICD-848P camera in light (light and dark field for NK) and luminescent modes (a set of light filters for excitation in the blue spectral region for the FDA). Color variables were averaged and measured in the Adobe Photoshop CS2 graphics editor. The analysis of the results was hampered by the presence of weakly stained organisms in the sample (19 out of 108 studied copepods), so discriminant analysis was required (StatSoft 8ΤΑΉ8ΤΙΟΑ 5.0). To construct the discriminant model, 5 independent variables (HSB, RGB) were used, and the classes “Live” (L) and “Dead” (D) were used as a training sample. The Brightness variable was excluded from the analysis as insignificant for classification (the data format is shown in Fig. 4). The resulting discriminant model is statistically significant (p <0.001). Information on the contribution of each independent variable to class distinction is presented in FIG. 5. Only two variables Green and Red turned out to be statistically significant (p <0.01) and made a decisive contribution to distinguishing classes. This is reflected in 2-parameter diagrams in which one of the variables is green. If the second variable is Blue (Fig. 6, A), the value of which in the discriminant model is small (p = 0.84), then the overlap of the L and D clusters is clearly noticeable. The color shade of Hue made a greater contribution to the classification (p = 0 , 26) and, together with Green, gave better cluster isolation (Fig. 6, B). Application of both statistically significant variables Green and Red in the diagram of FIG. 6B allowed to maximize the distance between clusters L and D.

The model classified all organisms with a controversial color, attributing them with some degree of probability to one of the classes - L (13 individuals) or D (6 individuals). A graphical representation of this classification is shown in FIG. 6, B. A fragment of the table of posterior probabilities of organisms belonging to classes is shown in FIG. 7. Only one individual from the training sample was marked in the table as being classified incorrectly (in Fig. 6, B it is indicated by an arrow). The discriminant model assigned it to class L, although by visual signs it was defined in class D. In this case, the copepod was well colored (which is reflected in its color characteristics), but it was classified by the researcher as dead because it had visual signs of decomposition. The discriminant analysis revealed this contradiction.

The inventive method has several advantages:

- the method allows to avoid subjectivity in the classification of organisms by the nature of their coloration, thereby increasing the accuracy and reliability of the results;

- the method is universal, since it can be used to solve other research problems in which an objective classification of organisms by degree of coloration is required.

Information sources:

1. Kastalskaya-Karzinkina M.A. Methodology for determining living and dead components of plankton on a fixed material // Transactions of the Limnological Station in Kosin. - 1935. - No. 19. - S. 91 - 103.

2. Gubareva E.S., Svetlichny L.S., Ishinilibir M., Belmonte G. Distribution of living and dead mesozooplankton in the bosporus regions of the Black and Marmara Seas: salinity tolerance of Acartia clausi and A. tonsa II Marine Ecological Journal. - 2008. - T. 7, No. 4. - S. 27 - 39.

3. Dressel D.M., Heinle D.R., Grote M.S. Vital starting to sort dead and live copepods // Chesapeake Sci. - 1972. - V. 13. - P. 156 - 159.

Claims (1)

A method for identifying living and dead organisms of mesozooplankton in marine samples, including sampling mesozooplankton, staining organisms with appropriate dyes, visual assessment of the color intensity of individuals under a microscope, characterized in that the visual assessment of the color intensity of individuals is combined with simultaneous microphotography of organisms using manual camera settings , keeping these settings unchanged during photography of at least one sample, after which the resulting images, using a raster graphics editor, for example, the Adobe Photoshop software package, measure average color and brightness characteristics for each individual and classify individuals as living or dead, performing a discriminant analysis of the measured digital values.

Images