A COMPARATIVE STUDY ON THE TOXICITY OF

THREE COMMERCIAL DETERGENTS ON

OREOCHROMIS NILOTICUS
Emilyn Q. Espiritu and Andrea B. Teran
Environmental Science Program, Ateneo de Manila University, Quezon City

Abstract
Ninety-six-hour toxicity tests were conducted on the three commercial detergents (Brands
B, T, and V) using tilapia (Oreochromis niloticus) as test organism. Test set-ups were monitored
daily for significant water parameters such as dissolved oxygen and temperature, as well as fish
mortality. Reference tests were also performed simultaneously using copper sulfate
(CuSO4· 5H2O). Tests were considered valid if the controls exhibited a ≥ 90% survival. The
results of the definitive tests indicate that the mean LC50s of the three detergents belong to the same
order of magnitude. Their rank order of toxicity is as follows: Brand B (mean 96-hr LC50= 12.04
± 1.22 mg/L) > Brand T (mean 96-hr LC50= 19.08 ± 5.93 mg/L) > Brand V (mean 96-hr
LC50= 41.88 ± 10.81 mg/L). The data indicate that Brands B and T are two to three times more
toxic than Brand V. Using a toxicity classification proposed by Espiritu (1994), the resulting LC50
values fall under the category of MODERATE TOXICITY. Given the fact that these three
detergents comprise nearly 28% of the total volume consumed by the market, their potential
adverse impact on our aquatic ecosystems must be thoroughly investigated.
Keywords: fish toxicity tests; LC50; detergents; copper sulfate; Tilapia; Oreochromis niloticus; water
quality

Introduction
Detergents are among the most important pollutants of the aquatic environment. The per
capita consumption of soaps and detergents in the Philippines was estimated to be 2.5 to 3.4 kg per
year in 1989 (1). Although many studies have been conducted on the toxicity of surfactants on
aquatic organisms, little is known about the impacts of detergents as complex chemical mixtures. In
a study done by Conti, the acute toxicity of three detergents on the lugworm (Arenicola marina)
exhibited 48-hour LC50 values from 12 to 15 mg/L (2). In another experiment done by Lal, et al.
using the water flea (Daphnia magna), median tolerance limit for a synthetic detergent was recorded
at 0.013 mg/L. The rate of survival of the organism decreased with increasing concentration and
time in a dose-dependent response (3). In the Philippines, a study by Dayrit and Espiritu determined
the toxicity of coconut fatty alcohol sulfonates (CFAS) and linear alkylbenzene sulfonates (LAS) to
three local freshwater fish species. Their results for Oreochromis niloticus showed mean 96-hour LC50
values for LAS and CFAS to be 8.06 ± 3.12 mg/L and 3.06 ± 0.92 mg/L, respectively (4). This
shows that CFAS is approximately 2.5 times more toxic than LAS, even though it proved to be very
biodegradable (4, 5).
However, there is an obvious lack of studies on detergents themselves in the country. It is
important that data on environmental effects such as toxicity be established, considering factors such
as temperature, dissolved oxygen and the type of test organism. The study intends to examine the
96-hour acute toxicity of three commercially available detergents on tilapia (Oreochromis niloticus), a
commercially important local fish species. The selection of the detergent products was based on their
market share by volume, as the three detergents chosen comprise nearly 28% of the market. These
were Brand B (ca. 10% national market share by volume), Brand T (ca. 17% national market share
by volume) and Brand V (<1% national market share by volume). Simultaneous toxicity tests will be
conducted with the reference toxicant, copper sulfate, CuSO4· 5H2O.

Materials and Methods

To date, there are no established standard toxicity test procedures in the Philippines for
determining the potential hazards caused by chemicals and various substances on the environment.
The methodology described in this paper was adopted from the toxicity test procedure developed by
Espiritu (6) at the Environmental Science Program of the Ateneo de Manila University.
Collection of Test Organisms. Juveniles of Oreochromis niloticus (approximately 1.5-2.0 cm) were
obtained from the Angat Gift Super Tilapia Hatchery Farm in Angat, Bulacan. They were placed in
polyethylene bags initially filled with oxygen to reduce fish mortality during transport.
Acclimation and Handling Procedure. In the laboratory, the polyethylene bags containing the fish
were initially placed in a 200-L acclimation tank filled with filtered and aerated tap water to about
70% volume. This allowed the water temperature inside the bag to equilibrate with the acclimation
tank temperature, reducing the effect of thermal stress on the fish. Two aerators were attached to
maintain dissolved oxygen levels at 7-8 mg/L. Each batch of fingerlings was acclimated for a period
of five to seven days prior to use in toxicity tests.
Toxicity Tests. Copper sulfate, CuSO4· 5H2O (Ajax Chemicals, Analytical Grade), was used as
reference toxicant in both range-finding and definitive toxicity tests. Stock solutions of the three
detergents and of copper sulfate were prepared using distilled water, while toxicant dilutions were
prepared using fresh filtered and aerated tap water. For the range finding tests with detergents, the
test concentrations differed by one order of magnitude, ranging from 0.1 to 1000 mg/L. For the
reference tests with copper sulfate, values ranged from 0.01 to 100 mg/L. The results were expressed
as a No Observable Effect Concentration (NOEC) and Lowest Observable Effect Concentration
(LOEC). For the definitive tests, the range of toxicant concentrations for both detergent and copper
sulfate were prepared following a logarithmic scale (7). All test set-ups were prepared in triplicate.
Each test aquarium was filled with 3 L of the respective test concentration. Using a fish net,
the fish were transferred about twenty at a time from the acclimation tank into a transfer aquarium.
They were then allowed to settle for an hour prior to transfer to the test aquaria in order to prevent
undue stress. Subsequently, an average of ten fishes were transferred from the transfer aquarium to
each test aquarium, starting from the control until the highest concentration. This procedure
prevented the contamination of the test solutions. The time of transfer was noted to ensure that
parameter measurements were done every 24 h. Toxicant concentrations were kept relatively
constant by replacing the solutions daily at approximately 70% volume. Human presence was kept
to a minimum in order to prevent unnecessary stress on the fish.
The fish were exposed for 96 h and were observed daily for signs of stress. Dead fish were
removed from each aquarium to prevent cannibalism and/or decomposition from stressing the
remaining test organisms. Dissolved oxygen and temperature levels were also monitored daily. No
feeding and aeration was done during the entire length of the test period except when dissolved
oxygen levels fell below 40%.
All assays were considered valid if the controls exhibited a ≥ 90% survival. Mortality was
assessed as the failure of the test organism to exhibit any form of movement upon prodding or
tapping of the test container within a 10-second observation period. The use of reference toxicity
tests provides experimental control, as it determines the conditions of the test organisms during the
bioassay. If reference toxicant test results are similar to the established norm, and control survival is ≥
90%, then mortality in the test organisms can be attributed to the toxicants tested. Lethality results
were expressed as 96-hour LC50s, i.e., the concentration of the detergent solution that kills 50% of the
organisms after 96 h of exposure. Definitive tests were repeated at least twice for purposes of
statistical validity.
Data Treatment. LC50 values were calculated using the trimmed Spearman-Karber Method
developed by Hamilton, et al. (1977) (8). These were subjected to simple linear regression to
determine the slope and establish the mode of action of the toxicant in function of time.
Mortality data were translated into percentages and subjected to Dunnett’s Test to determine
significant differences between the various test concentrations and the control for each detergent.
One-way ANOVA was also used to determine significant differences in the % mortality results
between the different detergents in function of time and concentration. Statistical tests were
performed using Toxstat 3.0 developed by Gulley, et al. and the USEPA (9). Simple linear regression
was used to determine the slope in each toxicant. These were compared to each other in order to
come up with a rank order of toxicity among the three detergents.

Results and Discussion

Results of Range-Finding Toxicity Tests. Table 1 shows the results of the range-finding toxicity tests for
the reference toxicant, copper sulfate, and the three detergents. The results for the reference toxicant
indicate NOEC and LOEC values of 1.0 and 10 mg/L, respectively. Moreover, the NOEC and
LOEC values for the three detergents are similar, ranging from 10 to 100 mg/L.
Results of Reference Toxicity Tests with Copper Sulfate. Table 2 shows the results of the definitive
tests with the reference toxicant expressed as LC50s in function of time. No LC50 value was derived
for the 24-hr period because the highest mean % mortality obtained was only 8.20% at the highest
concentration tested (10mg/L). After 48 h on the other hand, only one of the replicate tests
exhibited 50% mortality at the highest toxicant concentration. The resulting LC50 was found to be
10 mg/L. However, the 95% confidence limits were not reliable since the highest % mortality
obtained was only 50% at the 10-mg/L-concentration. At the end of the 96-hour test period, the
resulting mean 96-hour LC50 of copper sulfate to Oreochromis niloticus was 7.64 ± 0.75 mg/L. In a
study by Dayrit and Espiritu however, the same species responded more sensitively to copper sulfate,
yielding a 96-hour LC50 of 3.35 ± 1.40 mg/L, a difference of almost a factor of two (4). These values
of the LC50, however, still fall within the same order of magnitude (1-10 mg/L).

Table 1. Results of the toxicity tests for the three detergents and copper sulfate using Oreochromis niloticus.
(NOEC—No Observable Effect Concentration; LOEC—Lowest Observable Effect Concentration. n=3.)
Reference
Detergents
Tests
Parameters
Copper
Brand B Brand T Brand V
sulfate
Critical Range
1.0 – 10 10 - 100 10 - 100 10 - 100
(mg/L)
NOEC (mg/L) 1.0 10 10 10
LOEC (mg/L) 10 100 100 100
Table 2. Results of definitive toxicity tests with copper sulfate, using Oreochromis niloticus. Data expressed
as Mean LC50 (mg/L) ± SD; n=5; Trimmed Spearman-Karber method.
Exposure Period (hours) Mean LC50 ± SD (mg/L)
24 No LC50 derived
48 10.00 (limits not reliable)
72 8.25 ± 0.58
96 7.64 ± 0.75

Figure 1A shows the average mortality (in %) for tilapia exposed to the reference toxicant.
The data exhibit the typical dose-response, with mortality increasing with increasing concentration of
the toxicant and prolonged exposure period. Indeed, the average mortality increased from 8.5% after
24 h to 38% after 48 h, to 69.5% at 72 h and 73.2% at the end of the 96-hour exposure.
Figure 1B shows the trend of the mean LC50 values for copper sulfate through time. The
equation of the line presented offers an idea of how the toxicant behaves through time. A slope of –
1.18 indicates that the LC50 of copper sulfate decreases every 24 h at a rate of 1.18 mg/L.

Figure 1B. Mean LC50 trend through time for Oreochromis niloticus exposed to the reference toxicant
copper sulfate (n=5); NC = LC50 Not Calculable.

Table 3. Results of the definitive toxicity tests with the three detergents, using Oreochromis niloticus. Data
expressed as Mean LC50 (mg/L) ± SD; n=5.
Exposure Mean LC50 ± SD (mg/L)
Period
(hours) Brand B Brand T Brand V
24 13.27 ± 1.07 19.56 ± 6.50 68.12 ± 7.89
48 13.21 ± 1.19 19.56 ± 6.50 54.88 ± 13.15
72 12.48 ± 0.83 19.30 ± 6.17 47.57 ± 13.24
96 12.04 ± 1.22 19.08 ± 5.93 41.88 ± 10.81

Results of the Toxicity Tests with the Three Detergents. The results of the definitive toxicity tests for the
three detergents are presented in Table 3. The data are expressed as average LC50s in function of
time. The mean LC50 values for Brand B ranged from a minimum of 12.04 ± 1.22 mg/L to a
maximum of 13.27 ± 1.07, a difference of a factor of 1.1. Brand T LC50 values ranged from 19.08 ±
5.93 to 19.56 ± 6.50 mg/L, a factor difference of 1; while Brand V ranged from 41.88 ± 10.81 to
68.12 ± 7.89 mg/L, a factor difference of 1.63.
Detergent Brand B. Figure 2A shows the mean % mortality of Oreochromis niloticus exposed to
detergent Brand B. The results of the one-way ANOVA and Dunnett’s test (α= 0.05) indicate
significant differences in average percent mortality between the control and toxicant concentrations
starting from 10 mg/L at 24 h. A sharp increase in fish mortality was observed starting at 95% at a
concentration of 18 mg/L to 100% at the 100-mg/L concentration throughout the 96-hour test
period.
Figure 2B shows the LC50 trend of Brand B over the 96-hour period. The resulting
regression plot of the mean LC50s is almost constant throughout the 96-hour exposure period (See
also Table 3). This is also indicated in the low value for the slope, implying that the LC50 of Brand B
decreases by 0.442 mg/L every 24 h.

Figure 2B. Mean LC50 trend through time for Oreochromis niloticus exposed to detergent Brand B (n=5).

Detergent Brand T. Figure 3A shows the mean % mortality of Oreochromis niloticus to detergent
Brand T. The results of the one-way ANOVA and Dunnett’s test (α= 0.05) indicate significant
differences between the control and toxicant concentrations starting at 18 mg/L at 24 h. A sharp
increase in % mortality values was observed between the 10-mg/L and the 18-mg/L results, with the
latter averaging at 50% mortality at 96 h. As the toxicant concentration increased, the mortality also
increased from 95% at the 32-mg/L toxicant concentration after 24 h of exposure to 100% mortality
at the 100-mg/L concentration after 96 h.
Figure 3B shows the trend in the LC50 of Brand T to Oreochromis niloticus over a 96-hour
period. As can be seen, there is not much change in the mean LC50 over time, indicating that the
toxicity of Brand T remains constant throughout the length of the exposure period (See Table 3).
Indeed, the given slope is observed to be even smaller than that of Brand B, indicating minimal
change, a decrease of 0.17 mg/L per 24 h.

Figure 3B. Mean LC50 trend through time for Oreochromis niloticus exposed to detergent Brand T (n=6).

Detergent Brand V. Figure 4A shows the mean % mortality for Oreochromis niloticus exposed to
Brand V. The data show significant differences (one-way ANOVA and Dunnett’s test, α= 0.05)
between the control and the toxicant after 24 h exposure at the 56-mg/L-concentration. At the 10
and 18 mg/L concentrations average mortality remained at a constant value of 0% and 3.33%,
respectively, for 96 h. However, mortality increased with increasing exposure particularly at the 56-
mg/L-concentration, where mortality was 25.06% after 24 h, and peaked to 82.1% after 96 h. At
the highest toxicant concentration (100 mg/L), average mortality ranged from 95.2% to 98.3%
throughout the exposure period.
Unlike Brand B and Brand T, the LC50 values of Brand V considerably decreased with
increasing length of exposure, as shown in Figure 4B. The slope in the equation of the line given
emphasizes this difference, indicating that the LC50 value decreases by 8.603 mg/L every 24 h (See
also Table 3).

Figure 4B. Mean LC50 trend through time for Oreochromis niloticus exposed to detergent Brand V (n=6).

Figure 1A. Results of the definitive tests with reference toxicant copper sulfate using Oreochromis niloticus.
Data expressed as mean % mortality in function of concentration and across time (n=5).
Figure 2A. Results of the definitive tests with detergent Brand B using Oreochromis niloticus. Data
expressed as mean % mortality in function of concentration and across time (n=3-6). *significantly
different from control, α=0.05.
Figure 3A. Results of the definitive tests with detergent Brand T using Oreochromis niloticus. Data
expressed as mean % mortality in function of concentration and across time; n=6. *significantly different
from control, α=0.05.

Figure 4A. Results of the definitive tests with detergent Brand V using Oreochromis niloticus. Data
expressed as mean % mortality in function of concentration across time; n=6. *significantly different from
control.
Comparison of the LC50 Values Among Detergents. Figure 5 shows the regression plots for the three
detergents based on mean 96-hour % mortality values in function of concentration. Simple trends
analyses of the values reveal that there is not much difference in the slopes of the three detergents
indicating that there is a relatively proportional change in % mortality for every unit change in
concentration. However, the graph also shows that a rank order of toxicity can be deduced from the
location of the trend lines in the plot. As can be seen, Brand B can be considered the most toxic of
the three detergents, with its trend line closest to the y-axis, indicating % mortality of more than
20% at the 5.6 mg/L-concentration. This is followed by Brand T, which elicits 20% mortality at
around 10 mg/L, and then by Brand V, the least toxic, hitting the 20% mortality mark at
concentrations above 18 mg/L.

Figure 5. Regression plots of the mean % mortality values of the three detergents across toxicant
concentrations (n=5-6).

To check the validity and reproducibility of the results obtained in the various toxicity tests,
the coefficients of variation were calculated using the LC50 values for the three detergents. The data is
presented in Table 4. The results show that the coefficients of variation ranged from a minimum of
6.61% to a maximum of 10.15 for Brand B, from 31.10% to 33.21% for Brand T and from 11.58%
to 25.81% for Brand V. While it is desirable for test organisms to have a uniform response to a
toxicant, the inherent variability in the responses of a test organism to various stressors is generally
acknowledged. Such variations may be attributed to differences in the physiological state of various
batches of the test organism utilized in the toxicity tests (10).
Table 4. The coefficients of variation derived from LC50 data for each detergent across time. Data
expressed in %; n=5-6.
Exposure Period Brand B Brand T Brand V
(hours) (%) (%) (%)
24 8.04 33.24 11.58
48 9.02 33.24 23.96
72 6.61 31.99 27.84
96 10.15 31.10 25.81

In general, Brands B and T are two to three times more toxic than Brand V, although the
resulting mean 96-hour LC50 values fall within the same order of magnitude. Unfortunately, the
Department of Environment and Natural Resources has not yet come up with a system for toxicity
classification of detergent samples. Moreover, there are no studies in the literature that address the
toxicity of detergents in the context of a tropical environment. Such constraints provide limited
possibility by which comparisons with the results of the present study can be made.
However, Espiritu (11) proposed a system of classification in a study of 34 detergent samples
based on the resulting LC50s. This classification scheme was adopted in the present study with a
slight modification to quantify the number of assays to be “75% of the assays” instead of the
descriptive phrase “most of the assays.”
• Non-toxic = samples with LC50s > 1,000 mg/L in all assays;
• Low toxicity = samples with LC50s > 100 mg/L in 75% of the assays;
• Moderate toxicity = samples with 10 mg/L < LC50 < 100 mg/L in 75% of the assays; and
• High Toxicity = samples with LC50s < 10 mg/L in 75% of the assays.
Using this scheme in the present study results in a toxicity classification of “MODERATE
TOXICITY” for the three detergents tested. It must be noted, however, that such an arbitrary
assignment of values to come up with a toxicity scale must be approached with caution. From a
legislative point of view, the adoption of such a system of classification may lead to the formulation
of policies that may be too severe and overly-protective on one hand, or too lax on the other. Clearly,
more research is needed to validate this proposed scheme.
Notwithstanding the limited information that is available, certain facts can be derived. A
study by Chio et al. showed that the per capita usage of detergents in the Philippines in 1989 is at 2.5
to 3.4 kg per year (1). This translates to about 175 million kg of detergents used and dumped
directly into our waterways each year. The three detergents used in the present study comprise nearly
28% of the market share by volume, about 49 million kg per year. A rough approximation of the
detergent manufacturer’s suggestion for an effective cleaning solution indicates that for every 1 to 1.5
kg of clothes, a detergent concentration of 3,000 to 4,000 mg/L is needed. The results of the present
study showed that a minimum of 13.27 mg/L to a maximum of 68.12 mg/L of any of the three
detergents is sufficient to kill half of the fish after a 24-hour exposure. At the end of the 96-hour
exposure, 100% mortality was obtained at a concentration of 100 mg/L, which is 30 to 45 times
lower in concentration than the manufacturer’s recommendation.
This knowledge brings to light the possible environmental hazard that detergents may
present to our receiving aquatic ecosystems. In the absence of an efficient wastewater treatment
facility, which is typical of the conditions in a developing country such as the Philippines, there is an
urgent need to establish the appropriate infrastructure and water quality management system to
protect our water resources. Efforts must be directed at developing standardized toxicity test
procedures for determining the potential hazards caused by detergents and other substances. In line
with this effort, the present study proposes the above-mentioned toxicity test procedure as a tool for
monitoring purposes.

Conclusions and Recommendations

The results obtained in the present study indicate that the three detergents are of
MODERATE TOXICITY to the test organism Oreochromis niloticus. Moreover, for both definitive
tests, Brand B and Brand T are two to three times more toxic to the test organism than Brand V,
although the three detergents’ mean 96-hour LC50 values fall within the same order of magnitude
(10-100 mg/L). The results also show that the toxicity of Brand B and Brand T more or less remains
constant throughout the 96-hour test period; while the toxicity of Brand V increases with prolonged
exposure to the detergent.
Inasmuch as this study is geared towards the establishment of the toxicity of certain common
household products to local fish or other species in local environmental conditions, it is
recommended that: (1) chronic toxicity tests be done to firmly establish and document the toxicity of
the three chosen detergents on Oreochromis niloticus; (2) the test procedure and experimental design
be extended to include a wider range of detergents and other possible toxicants, especially those that
are widely-used, and whose effects are not yet documented; (3) toxicity tests also be done on the
specific components of such detergents, taking into account possible additive, synergistic,
antagonistic, or inhibitive action of the other components on the toxicity of another; and lastly (4)
that the test procedure extend to other local fish species, in order to contribute to the standardization
of fish toxicity tests in the country. It is hoped that with this study and a battery of other toxicity
tests, the Philippines can establish a vast documentation of products and their specific impacts on the
Philippine environment, using local test organisms.

References
1. J. T. Chio, E. T. Kua, F. M. Dayrit, Philippine Journal of Science, 1- 51 (December 1992).
2. E. Conti, Aquatic Toxicology 10, 325-334 (1987).
3. H. Lal, V. Misra, P. N. Viswanathan, C. R. Krishna Murti, Ecotoxicology and Environmental
Safety 8, 447-450 (1984).
4. F. M. Dayrit, E. Q. Espiritu, paper presented to the Soap and Detergent Association of the
Philippines, Makati, Philippines, February 2000.
5. T. W. La Point, Environmental Toxicology and Chemistry 12, 1749 (1993).
6. E. Q. Espiritu, unpublished data.
7. P. Douduroff, et al,. Sewage and Industrial Waste 23, 1380-1397 (1951).
8. R. A. Hamilton, R. C. Russo, R. V. Thurston, Environmental Science and Technology 715, 714-
719 (1977).
9. D. D. Gulley, A. M. Boelter, H. L. Bergman, in Toxstat Release 3.0, (University of Wyoming,
WY, 1989).
10. M. H. Depledge, Ambio 19, 251-252 (1990).
11. E. Q. Espiritu, doctoral dissertation, University of Gent, Belgium (1994).
Acknowledgements
The authors would like to acknowledge the Ateneo de Manila University, for research
funding; Mr. Billy Martin Gonzales for his assistance; the faculty and staff of the Environmental
Science Program for their support.