Introduction
Degradation in the water quality causes deterioration of stream ecosystems (Malmqvist & Rundle Reference Malmqvist and Rundle2002, Allan & Castillo Reference Allan and Castillo2007). Water quality degradation in streams is caused by anthropogenic impacts in the catchment, such as urbanization, industrial activities, and agricultural land conversion (Uriarte et al. Reference Uriarte, Yackulic, Lim and Arce-Nazario2011, Miller and Hutchins Reference Miller and Hutchins2017). For example, increased nutrient concentrations due to wastewater inflow from the catchment typically result in a loss of taxonomic diversity of stream-dwelling organisms through the disappearance of intolerant taxa (Lenat & Crawford Reference Lenat and Crawford1994, Roy et al. Reference Roy, Rosemond, Paul, Leigh and Wallace2003). Therefore, water quality management is of great importance for the conservation of biodiversity in stream ecosystems (Gasith & Resh Reference Gasith and Resh1999, Budi Prakoso et al. Reference Budi Prakoso, Fukusaki, Ueda and Miyake2022).
Benthic macroinvertebrates are a highly diverse taxonomic group in stream ecosystems, including sensitive taxa with varying degrees of tolerance to water quality gradients (Lenat Reference Lenat1988, Strayer Reference Strayer2006, Verdonschot & Moog Reference Verdonschot and Moog2006, Schletterer et al. Reference Schletterer, Füreder, Kuzovlev and Beketov2010). Therefore, the composition of the stream invertebrate community has been used as a comprehensive measure of water quality corresponding to the time scale of their life history (Karr Reference Karr1991, Bonada et al. Reference Bonada, Prat, Resh and Statzner2006). To date, a variety of biotic indices based on invertebrate composition have been developed to assess stream water quality worldwide (Hering et al. Reference Hering, Moog, Sandin and Verdonschot2004, Morse et al. Reference Morse, Bae, Munkhjargal, Sangpradub, Tanida, Vshivkova, Wang, Yang and Yule2007, Eriksen et al. Reference Eriksen, Brittain, Søli, Jacobsen, Goethals and Friberg2021). To calculate these invertebrate indices, scores are typically assigned to each indicator taxon according to its tolerance to poor water quality conditions, and then a community-level score is calculated to represent the water quality of a stream of interest (Hilsenhoff Reference Hilsenhoff1982, Ochieng et al. Reference Ochieng, Odong and Okot-Okumu2020).
Many invertebrate indices have been developed by improving and localizing the existing indices to suit the local environment and faunal composition of the target region that are often countries. For example, the classic biological monitoring working party (BMWP) scoring system was originally developed in the UK (Armitage et al Reference Armitage, Moss, Wright and Furse1983), and many localized versions have been derived in countries across Europe, North America, Latin America, Africa, and Asia (e.g. Czerniawska-Kusza Reference Czerniawska-Kusza2005, Nguyen et al. 2014, Ochieng et al. Reference Ochieng, Odong and Okot-Okumu2020). Similarly, the family biotic index (FBI) is an invertebrate index developed in the USA (Hilsenhoff Reference Hilsenhoff1982) and has been adapted in Europe, Africa, and Asia (Sekiranda et al. Reference Sekiranda, Okot-Okumu, Bugenyi, Ndawula and Gandhi2004, Morse et al. Reference Morse, Bae, Munkhjargal, Sangpradub, Tanida, Vshivkova, Wang, Yang and Yule2007, Etemi et al. Reference Etemi, Bytyçi, Ismaili, Fetoshi, Ymeri, Shala-Abazi, Muja-Bajraktari and Czikkely2020). Although the development of invertebrate indices through localization and their validation has been undertaken worldwide, such an approach has not been widely implemented in developing countries, typically in tropical regions where invertebrate diversity is generally high and severe water quality degradation is ongoing (Dudgeon Reference Dudgeon2000).
Indonesia is the largest tropical country with regard to population (ca. 278 million), and severe water pollution, primarily due to urbanization, has been reported, particularly on Java Island, where more than half of the population is concentrated (Ardania et al. Reference Ardania, Widyati, Rahmat and Machdi2022, Brontowiyono et al. Reference Brontowiyono, Asmara, Jana, Yulianto and Rahmawati2022). Some pilot studies have been conducted on the application and development of invertebrate indices in Indonesian streams. Dewi and Wardhana (Reference Dewi and Wardhana2020) and Muntalif et al. (Reference Muntalif, Chazanah, Ilmi, Sari and Bagaskara2023) applied current BMWP-related scoring systems to assess water quality. Furthermore, a new invertebrate index Biotilik has been developed for Indonesian streams based on FBI (Rini Reference Rini2011) and applied in some studies (Nugrahaningrum et al. Reference Nugrahaningrum, Harianja, Nugroho and Soesilohadi2017, Sandi et al. Reference Sandi, Arthana and Sari2017). However, to our knowledge, no study has compared the applicability of a wide range of indices or developed new indices based on solid empirical data, highlighting the need for studies that contribute to the development of reliable invertebrate indices for assessing water quality in Indonesian streams.
Herein, we compared the applicability of six existing invertebrate indices for assessing water quality in Indonesian streams. We used water quality and invertebrate data from a recent study conducted by Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023), which demonstrated the strong relationship between water quality and invertebrate assemblages in an environmental gradient typically observed in Indonesia. Then, we conducted TITAN to extract indicator taxa using the dataset and utilized the results to improve an existing index. Our ultimate goal was to develop and propose a better invertebrate index for assessing water quality that can contribute to water quality assessment and ecosystem management of Indonesian streams.
Methods
Invertebrate and water quality data
We used the invertebrate and water quality data from Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023). The data were collected at study sites established in 23 streams ranging from pristine forest streams to severely degraded urban streams in Yogyakarta Province, Java Island, Indonesia (Table S1 and Figure 1). The elevation of the study sites ranges from 53 m to 579 m. This area has typical Indonesian climatic characteristics, with temperatures ranging from 25 °C to 34 °C and annual rainfall of ca. 1,700 mm. We conducted the survey during the late rainy season (May 4–6, 2019; Lee Reference Lee2015), when streamflow is relatively stable and suitable for collecting representative invertebrate fauna (Wakhid et al. Reference Wakhid, Krisanti, Sumertajaya and Maryana2021; Harahap et al. Reference Harahap, Mahadewi, Ahmadi, Winoto, Ganiem, Rafikal and Hartanto2021). Benthic macroinvertebrates were collected by kick sampling using a D-frame net (0.5 mm mesh) at three locations (25 × 25 cm2 area) along the thalweg of riffles (Miyake et al. Reference Miyake, Makino and Fukusaki2021).
Figure 1. Location of study sites (white circles). See site codes for Table 1.
A surface water sample was concurrently taken at each site. In laboratory, invertebrates were sorted and identified to family level, if possible, using the taxonomic keys from previous local studies. Otherwise, they were identified to the lowest possible taxonomic level. The water samples were filtered and measured for nitrate (NO3-N), nitrite (NO2-N), ammonium (NH4-N), and phosphate (PO4-P) concentrations using an auto-analyzer (QuAAtro 2-HR) using a standard method (Budi Prakoso et al. Reference Budi Prakoso, Fukusaki, Ueda and Miyake2022). Subsequent analyses showed that among the 10 habitat parameters, invertebrate metrics had the strongest relationship with the PO4-P concentration in stream water (range: 2–166 µg l−1), which was typically high in severely urbanized streams (see Budi Prakoso et al. Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023 for more details). These results indicated the importance of the data collected by Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023) for evaluating the applicability of invertebrate indices for water quality assessment and the discriminative value of the PO4-P concentration as a proxy of stream water quality that affects invertebrate assemblages.
Existing invertebrate indices
We selected six existing invertebrate indices, BMWP UK (Armitage et al Reference Armitage, Moss, Wright and Furse1983), BMWP JPN (Nozaki Reference Nozaki2012), BMWP THAI (Mustow Reference Mustow2002), BMWP MY (Wan Mohd Hafezul Reference Wan Mohd Hafezul2016), SingScore (Blakely et al. Reference Blakely, Eikaas and Harding2014), and Biotilik (Rini Reference Rini2011), which could assess water quality in Indonesian streams. The selection criteria were as follows: the current dissemination of each index and the geographic proximity between the area where the index was developed and our target area (i.e. Yogyakarta Province, Indonesia).
The BMWP UK was developed in the UK; this index has a wide range of applications, and many BMWP-related procedures originated from the BMWP UK (Armitage et al Reference Armitage, Moss, Wright and Furse1983). The procedure assigns a score to the indicator taxa of water quality, ranging from 1 (most tolerant) to 10 (most sensitive). Then, the average score per taxon (ASPT, range: 1–10) was calculated as the average of the scores of each taxon found in the target stream to indicate water quality regardless of taxon richness. The BMWP JPN is an adaptation of the BMWP UK for Japanese rivers and streams, with the list of indicator taxa and the scores for each taxon modified to fit the Japanese invertebrate fauna. We selected the BMWP JPN because of its longest history of application among BMWP-related procedures in Asia, and this index has been modified since its utilization in the National Census on River Environment in 1993 (Nozaki Reference Nozaki2012).
The BMWP THAI and BMWP MY are also adaptations of the BMWP UK for streams in Thailand and Malaysia, respectively, which are developing countries in Southeast Asia (Mustow Reference Mustow2002, Zakaria and Mohamed Reference Zakaria and Mohamed2019). SingScore was developed for Singapore streams based on the macroinvertebrate community index (MCI), which was developed for New Zealand streams and is analogous to the BMWP (Blakely et al. Reference Blakely, Eikaas and Harding2014). A taxon score ranges from 1 to 10 similar to other BMWP-related indices, but the SingScore ranges from 20 to 200 multiplied by a constant of 20. Currently, Biotilik is the only invertebrate index developed for Indonesian streams. This index was a modification of the FBI, with indicator taxon scores ranging from 1 (most tolerant) to 4 (most sensitive) (Rini Reference Rini2011). The Biotilik score is calculated as the average of the taxon scores weighted by the relative abundance of each taxon (range: 1–4).
We calculated the scores of the abovementioned six indices using the full list of invertebrate taxa from Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023) for 23 study sites in Yogyakarta Province, Indonesia. We developed generalized linear models (GLMs) to examine the relationship between invertebrate indices and a water quality proxy, in order to assess the applicability of each invertebrate index as an indicator of water quality. The scores of each invertebrate index and the concentrations of PO4-P were used as response and explanatory variables, respectively. The error distribution was assumed to be Gaussian for all invertebrate indices. Statistical significance was set at P < 0.05.
Development of a new invertebrate index
We developed a new invertebrate index for Indonesian streams by modifying an existing index. The existing index modified was selected for its applicability as an indicator of water quality in our study sites (i.e. the highest coefficient of determination, R 2, by GLMs). We added new taxa that were not included in the target existing index as indicator taxa. The new indicator taxa were extracted using TITAN, which has been used in previous invertebrate studies to extract indicator taxa (e.g. Sarremejane et al. Reference Sarremejane, Stubbington, Dunbar, Westwood and England2019, Hu et al. Reference Hu, Zuo, Xu, Huang, Liu, Han and Bi2022). The TITAN combines indicator species analysis with nonparametric change-point analysis and uses indicator species scores to integrate the occurrence, abundance, and directionality of taxa responses along an environmental gradient (Baker and King Reference Baker and King2010). We targeted taxa that occurred at ≥5 stream sites (Baker and King Reference Baker and King2010) and were examined within a range of 2–166 µg l−1 of PO4-P concentration (Table S1). Bootstrapping (1000 replicates) was used to evaluate taxon-specific change points and distinguish negative (z−) and positive (z+) responses along the PO4-P concentration gradient. Only taxa with high purity and reliability (≥0.95) were considered indicators (Nguyen et al. Reference Nguyen, Boets, Lock, Forio, Van Echelpoel, Van Butsel, Utreras, Everaert, Granda, Hoang and Goethals2017). Then, we assigned a score to each extracted taxon based on the change point and its confidence interval along the PO4-P concentration gradient, with reference to the scores of other taxa that were originally included in the existing index. We retained the original taxon scores for taxa that were already included in the existing index.
We calculated the value of the new invertebrate index for each study site based on the procedure used in the target existing index. Analysis using GLM was performed in the same way as for the existing indices to compare the applicability of the new index for assessing water quality. All statistical analyses were performed using the R environment for statistical computing (version 3.6.3, R Development Core Team, 2020), with the glm2 package for GLMs and the TITAN2 package for TITAN (Marschner Reference Marschner2011; Baker et al. Reference Baker, King and Kahle2015).
Results
Relationships between existing invertebrate indices and water quality
A total of 13,268 individuals and 60 taxa of stream invertebrates were found in 69 samples collected from 23 study sites by Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023). The invertebrate abundance (N m−2) was on average 2,910 across the study sites (range: 336–17,765 N m−2) (Table S1). Each taxon was scored using the taxa list to calculate six existing invertebrate indices for each study site, if applicable (Table S2). Among the 60 taxa, 36 (60%) were used to calculate the BMWP THAI, SingScore, and Biotilik. The BMWP MY, BMWP JPN, and BMWP UK provided taxon scores for 33 (55%), 30 (50%), and 27 (45%) taxa, respectively. Six taxa (Acarina, Alpheidae, Gecarcinucidae, Podocopida, Psychidae, and Xiphocentridae) were found not to be included in the taxa list of any of the existing invertebrate indices.
The scores of six existing invertebrate indices varied among the 23 study sites. The BMWP index (possible range: 1–10) ranges were 3.00–6.38 for the UK, 5.00–7.57 for JPN, 3.00–6.36 for THAI, and 1.00–5.42 for MY (Table 1). The SingScore (possible range: 20–200) and Biotilik (possible range: 1–4) ranged from 32 to 133 and 1.00 to 2.50, respectively (Table 1). Analyses using GLMs revealed that all six existing invertebrate indices had a significant negative relationship with the PO4-P concentration, although the strength of the relationship differed among the indices (Figure 2). The BMWP THAI exhibited the strongest negative relationship with PO4-P concentration among six indices (R 2 = 0.471, P < 0.001), followed by the BMWP UK (R 2 = 0.432, P < 0.001). Contrarily, the relationship was relatively weaker for the SingScore (R 2 = 0.342, P = 0.003), BMWP MY (R 2 = 0.232, P = 0.020), BMWP JPN (R 2 = 0.208, P = 0.029), and Biotilik (R 2 = 0.176, P = 0.046), although the highest number of taxa (i.e. 36 taxa) was included to calculate the SingScore and Biotilik scores.
Table 1. Scores of the six existing indices and BMWP IDN for each study site. The location of the study sites is shown in Figure 1
Figure 2. Relationship between existing invertebrate indices and PO4-P concentration. The values inset into the graphs indicate the coefficients of determination (R 2) and P values obtained by generalized linear models (GLMs).
Development of a new Indonesian invertebrate index
TITAN revealed the community change along the PO4-P concentration gradient (Figure S1). Twenty-six taxa were found in more than five stream sites and then included in the analysis (Table S3). A total of 11 indicator taxa were identified using the indices of purity and reliability (Figure 3; Table S3). Among them, eight taxa showed a threshold decline (z−) in response to the increasing PO4-P concentration. The median of the change points in PO4-P concentration of the declining taxa, Euphaeidae, Ceratopogonidae, Baetidae, Heptageniidae, Ecnomidae, Psychomyiidae, Leptophlebiidae, and Elmidae, ranged from 7.5 to 62.5 µg l−1. By contrast, three taxa, namely, Podocopida, Oligochaeta, and Thiaridae, showed a positive response (z+) to the increasing PO4-P concentration. The median of the change points in PO4-P concentration of the increasing taxa ranged from 11.0 to 121.0 µg l−1.
Figure 3. Threshold indicator taxa analysis (TITAN) of invertebrate community response to PO4-P concentration. Horizontal lines represent 95% confidence intervals of observed change points (circles) for each taxon. Black circles and solid lines correspond to negative (z−) indicator taxa, whereas white circles and dashed lines correspond to positive (z+) indicator taxa. The size of change point symbol (circles) is proportional to the magnitude of the taxa response.
Of the 11 indicator taxa extracted by TITAN, five taxa did not appear to be included in the BMWP THAI scoring system. Then, we assigned taxon scores (possible range: 1–10) for five indicator taxa, namely, Euphaeidae, Ceratopogonidae, Ecnomidae, Elmidae, and Podocopida (Figure 3, Table S4). We assigned 10 to Euphaeidae and Ceratopogonidae because of their occurrence in a very low range of the PO4-P concentration gradient. Elmidae was assigned 7 because its change point and confident interval were slightly higher than those of Psychomyiidae (BMWP THAI score: 8). Ecnomidae was assigned 4 because its change point was similar to that of Baetidae (BMWP THAI score: 4) and the confidence interval was fairly inclusive. We assigned 1 to Podocopida because of its occurrence in a very high range of the PO4-P concentration gradient (Figure 3, Table S5).
The modified invertebrate index for assessing stream water quality in Indonesian streams (hereafter, BMWP IDN) was developed on the basis of the BMWP THAI by adding the five new indicator taxa. The BMWP IDN covered 41 taxa (68%) of the 60 taxa found by Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023) (Table S2). The BMWP IDN ranged from 3.00 to 6.92, which was wider than that of BMWP THAI (3.00 to 6.36). Analysis using GLM revealed that the BMWP IDN was significantly and negatively related to PO4-P concentration (R 2 = 0.487, P < 0.001; Figure 4). The relationship was stronger than any other existing invertebrate index examined in this study (R 2 < 0.471 for all; Figure 2).
Figure 4. Relationship between BMWP IDN and PO4-P concentration. The values inset into the graphs indicate the coefficients of determination (R 2) and P values obtained by generalized linear models (GLMs).
Discussion
In this study, we compared the applicability of six existing invertebrate indices using the dataset from 23 various Indonesian streams obtained by Budi Prakoso et al. (Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023) and found that BMWP THAI was the best index for assessing water quality in the target streams. Then, we developed a modified invertebrate index, BMWP IDN, by adding five new taxa extracted by TITAN to BMWP THAI by assigning taxon scores. Consequently, BMWP IDN showed a stronger relationship with water quality (PO4-P concentration) than any of the existing invertebrate indices. Some invertebrate indices such as BMWP UK and Biotilik have been applied to Indonesian streams in previous studies (Nugrahaningrum et al. Reference Nugrahaningrum, Harianja, Nugroho and Soesilohadi2017; Patang et al. Reference Patang, Soegianto and Hariyanto2018). Considering that these existing indices were fairly included in our analyses, our results implied the superiority of BMWP IDN at this stage. Therefore, this study can improve the invertebrate index to assess water quality and then promote the ecological conservation of Indonesian streams. We expect that our achievement would also facilitate the further development of invertebrate indices by incorporating the advanced methods, for example by incorporating multimetric indices (Hering et al. Reference Hering, Feld, Moog and Ofenboeck2006; Magbanua et al. Reference Magbanua, Hilario, Salluta, Alpecho, Mendoza and Lit2023), and by proposing a routine invertebrate monitoring in Indonesian streams that will enable long-term water quality assessment using invertebrate indices.
All of the six existing invertebrate indices showed significant negative relationships with PO4-P concentration, a proxy for water quality in our study area. This finding indicated the utility of each index as an indicator of water quality, but the strength of the relationship differed among the six indices. BMWP THAI has the closest relationship with water quality and can be characterized by the highest number of taxa that were included in the calculation of index values, along with the other indices developed in Southeast Asian countries, such as SingScore and Biotilik. In previous studies, the high number of indicator taxa included contributed to the high applicability of the invertebrate index for assessing water quality, and the number increased when the index was developed in more geographically proximate areas (Czerniawska-Kusza Reference Czerniawska-Kusza2005, Ochieng et al. Reference Ochieng, Odong and Okot-Okumu2020). Additionally, the relatively long trial history of BWMP THAI (Mustow Reference Mustow2002) compared with SingScore (Blakely et al. Reference Blakely, Eikaas and Harding2014) and Biotilik (Rini Reference Rini2011) is likely to contribute to its high utility. This supposition may be supported by the relatively strong relationship of BMWP UK with PO4-P concentration, which is the original of many BMWP-related procedures and has the longest history of application (Armitage et al Reference Armitage, Moss, Wright and Furse1983). Thus, our results suggest that the coverage of indicator taxa caused by geographical proximity and the historical persistence of indices determine the applicability of the invertebrate index when existing indices are applied to areas not yet well covered by those indices.
Herein, a new invertebrate index, BMWP IDN, was developed for assessing water quality in Indonesian stream. Using TITAN, we extracted 11 indicator taxa with negative and positive responses to the water quality proxy (PO4-P concentration), and five taxa were newly added to BMWP THAI to create the BMWP IDN by assigning taxon scores. Consequently, BMWP IDN successfully showed a wider range of scores across the 23 study sites and stronger relationship with water quality than any other existing invertebrate index. This improvement is primarily due to the fact that highly tolerant and intolerant taxa have been added to BMWP IDN scoring system. The addition of sensitive taxa to the environmental gradient has been reported to improve the applicability of invertebrate indices in previous studies (Mustow Reference Mustow2002; Fishar and Williams Reference Fishar and Williams2008). Notably, probably due to their small body size, Podocopida was not included in any of the existing invertebrate indices examined in this study, despite having the lowest taxon score in this study (highly tolerant taxon). The inclusion of such neglected but diverse and sensitive taxa may improve invertebrate indices (Mori and Meisch Reference Mori and Meisch2012). The addition of highly intolerant taxa, Euphaeidae and Ceratopogonidae, also contributed to sensitizing BMWP IDN to water quality gradients across our study sites (Aziz and Mohamed Reference Aziz and Mohamed2019, Rahman et al Reference Rahman, Burian, Creedy and Vogler2022). The new list of indicator taxa provided in this study can promote the future development of water quality assessments in Indonesian streams.
In conclusion, we proposed a new invertebrate index for assessing water quality in Indonesian streams. The extraction of uncovered sensitive taxa by TITAN can improve the existing index to sensitize the new index, BMWP IDN. Moreover, the extraction of the indicator taxa may have been facilitated by the selection of study sites (Budi Prakoso et al. Reference Budi Prakoso, Miyake, Ueda and Suryatmojo2023), which included different land use conditions and water quality. The coverage of a wide range of environmental gradients can contribute to the creation of a new invertebrate index (Mustow Reference Mustow2002; Haase and Nolte Reference Haase and Nolte2008). In this context, our new index, together with other possible ones, is expected to be applied and validated in other areas of Indonesia, and efforts should continue in the long term to develop a better invertebrate index for assessing water quality.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0266467424000257.
Acknowledgements
The authors are sincerely grateful to Muhammad Chrisna Satriagasa, Heni Puji Astuti, and Ghalbi Mahendra Putra for their assistance in the fieldwork. The Indonesian Directorate of Species and Genetic Biodiversity, Soekarno-Hatta Agricultural Quarantine Center, and Soekarno-Hatta Fish Quarantine Center provided the license to transfer invertebrates and water samples. Hidejiro Onishi kindly assisted with water quality analyses. We also would like to thank the members of the Stream Ecology Laboratory of Ehime University and the Watershed Management Laboratory of Universitas Gadjah Mada for their helpful advice.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The authors declare none.
Open access
