1. Introduction
Rivers and streams have complex ecosystem structures with a variety of environmental variables that make it impractical to assess all potentially important variables that influence various environmental conditions [
1]. Diatoms are sensitive to nutrient and organic matter concentrations and integrate long-term water quality with environmental variables [
2,
3]. Diatoms have traditionally been used to assess water quality; however, current research suggests that physical factors, such as habitat and landscape, may be linked to the organization of diatom assemblages in streams [
4,
5,
6,
7].
When the water becomes nutrient-saturated owing to the domination of similar eutrophic species, small differences in diatom composition and structure are influenced more by physical factors rather than nutrients [
8]. The degree to which species composition resembles that of the reference can be used as an indicator of biological circumstances, while deviations from the reference can more clearly explain the consequences of anthropogenic activity [
9]. The impact of nutrients and numerous environmental elements should be considered when evaluating aquatic ecological health as a whole. It is also vital to conduct research on water quality evaluation and river restoration technologies by employing diatom species that reflect the features of the region.
The biological assessment of water quality is critical for an overall perception of ecosystem integrity, which includes aquatic ecosystem habitat conditions [
10]. Bioindicators represent the impact of non-measurable environmental variables. Recently, epilithic diatom monitoring was used to assess biological water quality [
11]. The interactions between diatoms and environmental variables in streams are complicated, frequently including multiple variables that function in a hierarchical manner [
12]. Diatom autecology can benefit from explicitly incorporating interactions between variables. The current state of diatom autecology knowledge is insufficient and based on research that has not been explicitly targeted to assess common species’ environmental requirements. As a result, environmental preference lists for common species are frequently imprecise and inconsistent, and autecological information on common species is often unavailable in many places.
Classification and regression trees (CART) are effective for analyzing complex ecological data. A CART has the following characteristics: (1) the ability to utilize different types of response variables; (2) the ability for interactive exploration, description, and prediction; (3) invariance to transformations of explanatory variables; (4) simple graphical interpretation of complex results involving interactions; (5) model selection by cross-validation; and (6) procedures for dealing with missing values [
13]. Owing to its ability to handle both continuous and discrete variables, its inherent ability to model interactions among predictors, and its hierarchical structure, CART is appealing to many exploratory environmental and ecological studies. Regression trees (RT) and classification trees (CT) are useful for visualizing data structures and interactions [
13,
14]. A CART is well-suited for the relative abundance data of diatom species, as these frequently contain a large number of zero values [
15].
CART techniques are increasingly being utilized to investigate species–environmental interactions in plants and animals [
13,
16,
17]. Diatom autecology has been quantitatively evaluated using a weighted averaged regression model [
18,
19]. When a species is rare and has small ecological amplitude, and the distribution of the environmental variables among the sites is fairly homogeneous over the entire range of occurrence of the species, weighted averaging (WA) is just as effective as regression methods for estimating optima [
20].
The WA regression model remains a simple and useful tool used to reveal structures in data tables by rearranging species and sites based on an exploratory variable. WA has been proposed as a biotic index for vascular plants, algae, and faunal communities in streams and rivers [
21]. The WA regression model quantifies diatom autecology and, in many cases, expands our understanding of diatom species preferences. However, WA modeling assumes that the variable of interest is the only variable that influences species distribution. The significance of other environmental variables is implicitly accounted for in the calculation of the optimum WA. Conversely, WA cannot explicitly depict interactions between environmental variables. As a result, each environmental variable must be interpreted individually.
In Korea, the Trophic Diatom Index (TDI) [
22] is used to assess water quality. However, as each epilithic diatom species reacts differently under different environmental conditions, it is necessary to establish a method that can be used as an indicator of water quality evaluation by analyzing the habitat preferences of epilithic diatoms. Therefore, it is necessary to develop biological indicators and evaluation methods that are appropriate for domestic conditions and reflect the effects of various environmental factors. The CART and WA were used in this study to determine the quantitative parameters of epilithic diatom habitat preference based on physicochemical environmental factors. Most previous studies have focused on the epilithic diatom community. However, as diatoms respond to complex environmental variables in a sensitive manner, habitat preferences for diatom organisms should be qualitatively indicated. In particular, the dominant species is a species that has been well adapted to the region’s environmental conditions and defines the representative characteristics that characterize the group. Dominant species are those that appear at the highest density and frequency and are used for ecological river restoration as reflected in river planning, design, construction, maintenance, and monitoring through the characterization of the river environment [
23].
The goal of this study was to use CART and WA to determine the physicochemical environmental factors that influence the dominant diatom species as well as to determine the preferred habitat for each species. The results of this study can be used to evaluate and predict changes in the water environment.
4. Discussion
The complexity of hierarchical interactions between environmental variables and diatoms may contribute to the difficulty in linking indicator taxa to a specific set of environmental variables. Mismatches between diatoms and measured environmental variables at both spatial and temporal scales could also be a source of contention [
7].
The various water pollution features have an impact on epilithic diatom assemblage changes, which could result in a cascade of impacts that determine diatom assemblage composition.
Epilithic diatom species are sensitive to a wide range of physicochemical conditions and have complex interactions with their environment. The most prevalent and dominant species,
A. convergens,
A. minutissima,
N. amphibian,
N. fonticola, and
N. inconspicua, may be able to adapt to a variety of environmental changes. Typically, each species is found near specific environmental optima [
31].
A. minutissima was discovered primarily in well-protected areas and reference sites. The best water quality was found in
A. minutissima dominated areas, which were mostly located upstream [
32,
33].
N. palea [
29] was the most frequently encountered taxon, was primarily influenced by the chemical properties of water and lived in heavily polluted areas [
34].
N. palea, a nutrient-tolerant species, has been discovered in both intensively farmed agricultural areas and urban environments [
35]. In the CART analysis of the six dominant diatom species, 10 variables (AT, CP, Agc, Wd, temp, DO, Cdv, Chl_a, COD, TP, DTP, and DTN) were used multiple times for decision making, as is shown in
Table 5. The most commonly used variables in CT and RT were AT and Cdv (
Table 7).
CT indicates whether an epilithic diatom is present, while RT indicates the relative likelihood of how many epilithic diatoms are likely to appear. Our findings distinguished CT from RT. Although certain environmental variables were not necessarily the most influential, based on the aforementioned analyses, we could determine the impact of environmental variables.
Table 8 shows the change points of the CART and WA optima. CART denotes environmental variables with a changing point range, and the appearance ratio of the species has high values within the specified range, while WA denotes variables with a specific value, optima, which denotes the optimal condition for the species to appear.
The changing point range in CART for
A. minutissima was not identical to the WA optima in the COD and DTP variables. WA optima were included in the range of changing points in CART for
A. convergens. There was little difference between the range of changing points and WA optima for
N. inconspicua, TP for
N. amphibia, and DTN for
N. palea. The values of the range of changing points in the CART and WA optima were not identical for
N. fonticola (
Table 6).
CART and WA optima demonstrated that various physicochemical environmental factors, such as AT, Cdv, and nutrient concentration, influenced the occurrence and composition of epilithic diatoms. Changes in specific environmental variables may affect the abundance of species that prefer the same conditions.
The CART and WA optima results for the six dominant species (
A. minutissima,
A. convergens,
N. amphibian,
N. fonticola,
N. inconspicua, and
N. palea) revealed that each species was influenced by a variety of complex environmental variables. Furthermore, the range of the changing point and WA optima did not always correspond with each other owing to the upper–lower rank interaction. The CART and WA approaches yielded distinct yet complementary information on the complex relationships between common stream diatoms and environmental variables [
15].
A. minutissima and A. convergens were the dominant species in the reference condition in this study and were highly prevalent under the following conditions: conductivity <154 and DTP < 0.003, and altitudes >102 and width >152, respectively. As changing physical factors, such as altitude and width, which are the habitat conditions of A. convergens, to suit the habitat conditions of one type of diatom is difficult, the habitat environment of A. minutissima, which is primarily affected by chemical factors, should be considered. For example, if the conductivity and DTP in a specific river are changed to conductivity <154 and DTP < 0.003, the habitat conditions of A. minutissima, it is projected that the chance of maintaining good aquatic health would rise in tandem with the rate of the appearance of A. minutissima. However, more research and verification of the properties of organisms living in the same habitat are required before they can be used for river restoration.
The autecological properties of diatom species may be useful for objectively determining a reference condition. The environmental preferences of indicator diatoms are considered when developing river restoration policies and quantitative evaluation criteria for biological assessment.
Based on the findings of our study,
A. minutissima and
A. convergens, which had high dominant frequencies in groups C and D, respectively, belonged to the “excellent” and “good” classes in the TDI grade, while
N. inconspicua, which had high dominant frequencies in groups A and B, also belonged to the “excellent” and “good” classes in the TDI grade, appeared to be a member of the “fair” class. The main species in Groups A and B,
N. palea, belonged to the “poor” class and it was determined that the concentration of nutrients and the water conversion process were strongly associated (
Table 9).
According to the findings of this study, the abundance and composition of epilithic diatom species can be influenced not only by nutrient concentrations but also by a variety of physicochemical environmental factors. Furthermore, it was determined that even the frequency of occurrence of species living in the same habitat may change owing to changes in specific environmental factors.
5. Conclusions
Physical and chemical factors influence the composition of diatom assemblages. Diatom species respond differently to physicochemical variations and complex interactions among environmental variables. The autecological characteristics of diatom species may be useful for conducting an objective search for reference conditions. The environmental preferences of indicator diatoms should be considered when developing river restoration policies and quantitative evaluation criteria for biological assessment.
In this study, environmental factors influencing structural changes in epilithic diatom species were quantified. The composition and number of epilithic diatom species appeared to be influenced by a variety of complex environmental factors depending on the habitat. CART and WA analyses were performed on dominant species to quantitatively derive environmental factors affecting each species’ relative abundance and to present environmental conditions reflecting each species’ physiological and ecological characteristics.
The presence (0.75) of A. minutissima was determined as high under the CP ≥ 2.5% and DO < 16.9 mg/L conditions. The presence of A. convergens (0.58) was determined by the DTN < 4.218 mg/L condition. The presence (0.27) of N. inconspicua was determined by altitude < 90.5 m and CP ≥ 7.5%. The presence (0.50) of N. fonticola was determined by altitude < 134.5 m. The presence (0.64) of N. amphibia was determined by conductivity <120.5 μS/cm. The presence (0.64) of N. palea was determined by conductivity ≥172.5 μS/cm as well. CART analysis may help identify the hierarchical interactions among environmental variables in predicting the relative abundance of epilithic diatoms.
Research on the autecological characteristics and environmental preferences of indicator diatom species could aid in making objective decisions for the establishment of river restoration policies and quantitative evaluation criteria for biological assessments. In the future, it is expected that a clear set of integrated water quality evaluation criteria reflecting optimal diatom habitat conditions will be established. Basic data for the establishment of standards and the development of ecological river restoration technologies are expected to be available.