Microbiological and physicochemical characterization of olive mill wastewaters from a continuous olive mill in Northeastern Portugal

Available online at www.sciencedirect.com

Bioresource Technology 99 (2008) 7215–7223

Microbiological and physicochemical characterization of olive millwastewaters from a continuous olive mill in Northeastern Portugal

Carla Amaral a, Marco S. Lucas c, Joao Coutinho c, Antonio L. Crespı a,Maria do Rosario Anjos a, Celia Pais b,*

a Centro de Estudos Tecnologicos do Ambiente e da Vida (CETAV) – Departamento de Eng, Biologica Ambiental – UTAD, Quinta de Prados,

Apartado 1013, 5001-801 Vila Real, Portugalb Centro de Biologia (CB-UM) – Departamento de Biologia, Universidade do Minho, Campus de Gualtar, 4710-057 Braga, Portugal

c Centro de Quımica (CQ) – Departamento de Quımica – UTAD, Quinta de Prados, Apartado 1013, 5001-801 Vila Real, Portugal

Received 10 May 2007; received in revised form 18 December 2007; accepted 20 December 2007Available online 7 February 2008

Abstract

The microbiological and physicochemical characterization of samples from the different wastewaters generated during oil extractionin a continuous olive mill was performed. The main aim was to determine which of the physicochemical parameters were the best fitted tocorrectly characterize these residual waters. High correlations were obtained for COD, DOC, K, P and N contents with the samplingpoints, allowing the distinction of olive washing waters (OWW) from olive centrifuge waters (OCW) and olive mill wastewaters(OMW). These parameters were sufficient for a rapid and less costly chemical characterization of these waters. Phenols and oil and greasecontents, together with low pH and dissolved oxygen contents, and high organic loads, were the most toxic for microbial populations.Microbial characterization showed that fungi were well adapted to these stressing environmental characteristics and the reuse of OMWafter aerobic treatment with microbial species isolated from the effluent is considered.� 2008 Elsevier Ltd. All rights reserved.

Keywords: Olive mill wastewater; Physicochemical characterization; Heterotrophs; Filamentous fungi; Yeasts

1. Introduction

Olive mill wastewater (OMW) generated by the oilextraction industry is a very important pollutant in Medi-terranean countries, which are responsible for about 95%of the worldwide olive oil production (Fiorentino et al.,2003). This activity occurs between December and Febru-ary and the seasonal polluting load is mainly due to thehigh chemical oxygen demand (COD) values, together withthe generation of large quantities of water (Aktas et al.,2001). The high organic load in OMW has a toxic effectespecially on seed germination and methanogenic bacteria.Furthermore, the high polyphenol and fatty acid contentsof these waters can inhibit the growth of microorganisms

0960-8524/$ - see front matter � 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.biortech.2007.12.058

* Corresponding author. Tel.: +351 253 604 312; fax: +351 253 678980.E-mail address: [email protected] (C. Pais).

and stop conventional secondary and anaerobic treatmentsin municipal treatment plants (Tsioulpas et al., 2002; Casaet al., 2003; Sassi et al., 2006). Due to the importance ofOMW as pollutants their disposal constitutes a seriousproblem and the development of effective treatment tech-nologies remains a priority since OMW is still dischargeddirectly to the environment, without any treatment, inmany developing countries.

Several studies regarding the chemical characterizationof OMW are available but their results are very variabledue to factors which cannot be controlled, such as climaticconditions, olive cultivars, degree of fruit maturation, stor-age time and conditions, and oil extraction procedure(Aktas et al., 2001; Casa et al., 2003; D’Annibale et al.,2004; Fiorentino et al., 2003; Mouncif et al., 1993; Robleset al., 2000; Sassi et al., 2006; Vitolo et al., 1999). On theother hand, little is known about the microbial flora

7216 C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223

present in OMW. To our knowledge, the only reports arefrom Mouncif et al. (1993), referring to traditional millsin Morocco, and from Sassi et al. (2006), which comparethree olive oil extraction processes also in Morocco.

In the present work the microbiological and physicochem-ical characterization of the different types of water generatedduring the oil extraction chain in a continuous mill located inNortheastern Portugal was performed. The objectives of thisstudy were twofold: (1) to determine which of the physico-chemical parameters were the most suitable to be used to dis-tinguish among sampling points; and (2) to perform amicrobiological characterization in an attempt to evaluatethe effects of the different physical and chemical parameterson the microbial flora present along the extraction process.

2. Methods

2.1. Sampling

Samples were collected from a continuous olive milllocated in Northeastern Portugal, at different times, duringtwo campaign periods. Washing is the first step of oliveprocessing in the factory to clean the fruit from leafs, dustand small portions of garbage that may be adhered to theolive’s pellicle. During the course of this study the millunderwent some remodelations. As a result the washingprocess, which was performed inside the factory, wasreplaced by a modern one located outside the factory.Thus, during the 2003/2004 extraction, samples from theolive washing water were collected immediately after wash-ing, and during the 2004/2005 extraction period sampleswere collected from inside the machine, where they hadbeen stored for variable periods of time. In both campaignsthe sampling points were selected in order to obtain the dif-ferent wastewaters generated during oil extraction: olivewashing waters (OWW), olive centrifuge waters (OCW),and olive mill wastewaters (OMW). All samples were main-tained at 4 �C and analysed within 24 h after collection.

2.2. Microbiological parameters

For microbial count, 1 mL of each sample was diluted in10 mL of peptone water (0.1% w/v) and serial dilutionswere prepared to obtain CFU counts in the range of 30–300 per plate. 100 lL of the corresponding decimal dilu-tions were inoculated in three different media: Plate CountAgar (PCA – Difco), Potato Dextrose Agar (PDA – Difco)and Yeast Malt Agar (YMAgar – Difco). The plates wereincubated at 20 �C, 25 �C and 30 �C for a week, and thecolony forming units (CFU) counted. The mean valuesobtained with three replicas per sample, per medium andper incubation temperatures are presented.

2.3. Physicochemical parameters

For the physical and chemical analysis, the pH, dis-solved oxygen (DO), and temperature (Temp) were mea-

sured in situ after collection of the sample with amultiparameter analyser model 340i/SET from WTW,according to the manufacturer’s instructions. Chemicaloxygen demand (COD) was measured by molecularabsorption spectroscopy by using the potassium dichro-mate digestion method #8000 in a HACH 2400 analyser,and total phenols in mgcafeic acid/L were measured by molec-ular absorption spectroscopy by using the modified Folin–Ciocalteau method (Peres, 2001). Oil and grease and totalsolids were measured according to methods 5520 and2540, respectively, from APHA (1992). Dissolved organiccarbon (DOC), Kjeldhal nitrogen (NKj), and total phos-phorus (P), NO3–N (Nnit) and NHþ4 –N (Namon) were mea-sured by molecular absorption spectroscopy, by using forDOC the UV persulfate digestion, the sulphuric digestionmethod for NKj and P, the Bhertlot reaction for Namon,and cadmium reaction for Nnit. The content in K was mea-sured by using the flame emission spectroscopy methodafter sulphuric digestion. The values for organic nitrogen(Norg) resulted from the difference between measuredKjeldhal nitrogen (NKj) and measured ammoniac nitrogen(Namon). The results presented are means of threereplicates.

2.4. Statistical analysis

Microbiological counts were analysed by means ofANOVA tests by using the Excel Program from MicrosoftOffice Software. Physicochemical parameters, as well astheir interactions with microbiological parameters, andmultivariate analysis were performed and analysed usingthe STATISTICA 7 SOFTWARE for Windows.

3. Results and discussion

3.1. Analysis of microbial counts along the oil extraction

chain

Samples were collected from (i) olive washing waters(OWW), (ii) olive centrifuge waters (OCW), and (iii) olivemill wastewaters (OMW) in order to obtain a completemicrobiological characterization along the oil extractionprocess. Three samples from each extraction campaignwere analysed, one collected in the beginning, one in themiddle, and the other at the end of each campaign period.Microbial counts regarding total heterotrophs, filamentousfungi and yeasts were performed for each sample. Theplates were incubated at different temperatures, but incuba-tion temperatures did not affect significantly the CFUcounts of the microbial groups analysed in any of theselected sampling points. Thus, the results presented arethe ones obtained in the plates incubated at 25 �C.

As referred previously, the first step of olive processing inthe factory was washing. In the first sampling period (2003/2004 campaign), water samples were collected directly out-side the factory. In the second campaign (2004/2005) thesamples were collected from the water stored in the machine

C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223 7217

deposit. Microbial counts are presented in Table 1. Themean CFU counts obtained for total heterotrophs werealways higher than the ones obtained for filamentous fungiand for yeasts. These differences were more obvious inOWW than in the other sampling points. Generally, thepopulation of heterotrophs suffered a small decrease inthe second sampling date, recovering later on. The popula-tions of filamentous fungi and yeasts showed a similar evo-lution during both extraction periods and this was observedfor all sampling points. No significant variations werefound in microbial counts between sampling dates, duringthe two campaigns.

Comparing the mean values obtained for the differentmicrobial groups in the first sampling point (OWW) alongthe time, it is noteworthy that, for yeasts and heterotrophs,there were significant differences among sampling dates inboth campaigns. Furthermore, in the first sampling yearhighly significant statistical differences in filamentous fungicounts were observed, while in the second campaign thesewere not found. Higher mean values of total heterotrophs,moulds and yeasts could be observed in OWW during thesecond extraction campaign. These differences could berelated to the fact that in that period washing water wasstored in the deposit before the samples were collected,

Table 1Variation of mean CFU counts within and between the two extraction campaiand olive mill wastewaters

Sampling point Collection date Heterotrophs

Mean (CFU/mL) p value

OWW

(a) Begin (OWW1) 3.96 � 109

Middle (OWW2) 2.17 � 108 0.0000End (OWW3) 7.40 � 108 ***

(b) Begin (OWW4) 1.57 � 109

Middle (OWW5) 3.90 � 108 0.0462End (OWW6) 2.55 � 109 *

OCW

(a) Begin (OCW1) 1.15 � l07

Middle (OCW2) 4.03 � l06 0.2851End (OCW3) 1.69 � 107 ns

(b) Begin (OCW4) 2.93 � l06

Middle (OCW5) 5.50 � l06 0.3645End (OCW6) 2.03 � l06 ns

OMW

(a) Begin (OMW1) 1.68 � 108

Middle (OMW2) 1.14 � 107 0.0000End (OMW3) 6.10 � l06 ***

(b) Begin (OMW4) 9.98 � l05

Middle (OMW5) 7.63 � l06 0.3645End (OMW6) 6.56 � l06 ns

The statistical p values presented are calculated from analysis between collcampaign; ns: not significant.

* Significant for p < 0.05.** Very significant for p < 0.01.

*** Highly significant for p < 0.001.

while in the first campaign the samples were collectedimmediately after washing.

Samples collected from the vertical centrifuge (OCW)resulted from phase separation (aqueous and oily fractions)after grinding at about 37 �C. High numbers of all micro-bial populations were obtained in these samples duringboth extraction periods (Table 1). However, the hetero-trophs were mainly dominated by fungal colonies. Micro-bial populations isolated from OCW in the first campaignpresented similar values among the three groups. Compar-ing these numbers with those obtained for olive washingwaters (OWW) the populations suffered a marked decrease.This could be related with the physicochemical environ-ment, which was less favourable to microbial growth anddevelopment. In fact, the contents in solids and relatedparameters also increased significantly (Table 2). The pop-ulations of filamentous fungi and yeasts showed statisticalsignificant variation in numbers between the two cam-paigns and these differences might be again related withthe environmental characteristics present in the twocampaigns.

Samples designated by OMW were the final waters col-lected after a variable stabilization period in a small basinlocated inside the factory. Apparently the stabilization

gns in samples obtained from olive washing waters, olive centrifuge waters

Filamentous fungi Yeasts

Mean (CFU/mL) p value Mean (CFU/mL) p value

5.62 � 108 4.49 � 108

8.45 � l06 0.0002 6.16 � l06 0.00003.29 � 107 *** 3.11 � 107 ***

5.74 � 107 1.79 � 108

1.44 � 108 0.1796 5.56 � 107 0.00244.39 � 108 ns 3.63 � 108 **

1.48 � 107 7.78 � l05

3.77 � l06 0.0435 4.98 � l06 0.04592.05 � 107 * 8.15 � l06 *

1.31 � l06 6.06 � l05

3.54 � l06 0.1426 4.12 � l06 0.00191.36 � l06 ns 5.82 � l05 **

4.89 � l06 5.12 � l06

3.67 � l06 0.0047 1.66 � l06 0.29312.73 � l06 ** 1.63 � l06 ns

3.54 � l05 8.12 � l05

5.50 � l06 0.5325 4.35 � l06 0.48665.56 � l06 ns 4.14 � l06 ns

ection dates in each campaign. (a) 2003/2004 campaign; (b) 2004/2005

Table 2Physicochemical parameters measured in samples from OWW, OCW and OMW during the two campaign periods

Samplingpoint

Collectiondate

pH DO Temp COD BOD DOC Totalsolids

Oilandgrease

Totalphenols

K P NKj Nnit Namon Norg C/Nratio

OWW

(a) OWW1 6.6 6.1 13.9 1276 550 144.6 0.84 0.20 11.7 47.4 2.0 8.2 <0.01 0.83 7.37 18OWW2 6.4 6.1 8.8 1018 200 101.3 0.67 14.0 4.9 49.0 2.6 3.5 <0.01 0.87 2.63 29OWW3 6.8 5.9 12.8 988 300 419.4 0.68 5.93 9.1 35.1 2.0 2.6 0.04 0.82 1.78 161

(b) OWW4 5.5 6.8 12.7 1951 500 603.7 0.60 5.41 981.0 147.9 4.6 19.7 0.2 1.05 18.65 31OWW5 7.5 8.0 10.3 1780 200 323.3 0.60 2.50 1027.0 26.9 2.8 1.2 <0.01 0.29 0.91 269OWW6 5.2 8.0 5.0 – – 1018.9 0.95 0.40 1000.0 208.7 6.1 11.4 <0.01 1.47 9.93 89

OCW

(a) OCW1 5.7 – 23.3 17888 1000 1138.9 9.98 9.0 14.7 106.9 9.3 23.0 <0.01 0.80 22.2 50OCW2 5.4 7.9 22.0 76450 700 13311.7 51.9 55.6 82.0 1920.5 149.2 186.0 0.26 1.74 184.3 72OCW3 – – – 3725 0 764.8 1.49 9.1 7.1 22.0 1.8 5.5 <0.01 0.79 4.71 139

(b) OCW4 5.8 6.4 31.7 9510 2000 1157.9 4.31 16.0 853.0 232.9 22.9 40.7 0.28 0.83 39.87 28OCW5 6.2 6.4 29.5 13400 1500 2865.6 5.24 9.3 753.0 295.2 21.3 41.3 <0.01 1.80 39.5 69OCW6 6.2 5.3 24.2 – – 1398.5 36.9 38.65 1043.0 104.4 10.4 20.2 <0.01 1.53 18.67 69

OMW

(a) OMW1 5.3 0.2 25.7 18138 1500 3147.3 7.11 2.47 27.0 384.9 32.0 55.3 <0.01 0.84 54.46 57OMW2 5.2 0.9 24.4 18550 2000 4739.3 5.26 8.6 40.0 669.3 52.2 77.4 0.16 1.47 75.93 61OMW3 5.1 3.4 29.0 7450 500 1893.9 3.13 6.93 41.0 169.4 11.5 22.6 <0.01 1.01 21.59 84

(b) OMW4 5.2 2.9 23.6 28640 8000 13372.9 7.78 22.4 993.0 1586.5 115.6 102.4 0.28 0.99 101.4 130OMW5 5.8 0.2 26.3 68480 9500 17279.5 30.22 62.3 955.0 2210.5 162.5 302.7 <0.01 6.40 296.3 57OMW6 5.6 1.4 23.8 – – 5584.3 6.17 9.3 1051.0 581.0 43.9 52.6 <0.01 1.90 50.70 106

Data units: DO – mg O2/L; temp – �C; COD – mg O2/L; BOD – mg O2/L; DOC – mg C/L; total solids – g/L; oil and grease – g/L; phenols – mgcafeic acid/L;K – mg K/L; P – mg P/L; Ntot – mg N/L; Nnit – mg N/L; Namon mg N/L; Norg – mg N/L. (a) 2003/2004 campaign; (b) 2004/2005 campaign.

7218 C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223

period had no impact on the microbiota since the meancounts obtained for the populations of heterotrophs, fila-mentous fungi and yeasts in this sampling point were verysimilar to those reported for olive centrifuge waters (Table1). During the first extraction period (2003/2004) thecounts obtained for the microbial groups showed signifi-cant differences between sampling dates for filamentousfungi and heterotrophs, while no differences were observedfor yeast CFU counts. In the second sampling period(2004/2005) no significant statistical differences betweencollection dates could be found for any of the microbialgroups.

3.2. Analysis of physicochemical parameters

The results obtained for the several physicochemicalparameters measured in the three sampling points are sum-marized in Table 2.

In the first sampling point (OWW) a continuous drop inthe temperature values was observed during the secondsampling period. This was probably a consequence of thedecrease in environmental air temperatures, since at thattime the washing machines were located outside the factory.The washing waters were slightly acidic and had low DOlevels. The mean values for organic matter, measured asCOD and BOD were 1402 mg O2/L and 350 mg O2/L,respectively, and were much higher in what concerns

COD in the second sampling period. The COD/BOD ratiowas approximately 30%, meaning that this water, althoughnot heavily loaded, was not totally degraded by biologicalmeans. This was probably due to the presence of some bio-logical inhibitors for microbial aerobic growth. The pheno-lic content, measured as total phenols, showed a markedincrease in the second sampling year. In the 2003/2004 sam-ples, the mean value was 8.5 mgcafeic acid/L which increasedto 1003 mgcafeic acid/L in 2004/2005. This increase might berelated to a higher maturation degree of the olives in the sec-ond sampling periods. Besides, the polyphenols were posi-tively correlated with COD values (Table 3) and thosewere higher in the second sampling period.

The oil and grease fraction resulted from the liberationof oils from the olive fruit into the medium. The degreeof olive maturation and the integrity of the fruit influencethe liberation of oils and might explain the variable valuesobtained, mainly during the first extraction campaign. Thesolids measured were in agreement with the valuesobtained for COD and also with the ones for organic car-bon, measured as DOC. In the second campaign periodhigher values were obtained both for total solids and forDOC, but these variations showed no statistical signifi-cance. Apparently, these two parameters were directly cor-related with the carbon inherent to the olives and with thesolids washed from their surface. Therefore, these differ-ences between sampling years might be related with the

Table 3Pearson correlations in physicochemical parameters measured in samples from OWW, during the two campaign periods

OWW correlations N = 6 pH COD Total phenols K P NKj Namon Nnit Norg

pH 1.0COD ns 1.0Total phenols ns 0.95* 1.0K �0.91* ns ns 1.0P ns ns ns 0.90* 1.0NKj �0.89* ns ns 0.97** ns 1.0Namon �0.90* ns ns ns ns ns 1.0Nnit ns ns ns 0.96** 0.90* 0.91* ns 1.0Norg �0.88* ns ns 0.97** ns 1.00*** ns 0.91* 1.0

ns: Not significant.* Significant for p < 0.05.

** Very significant for p < 0.01.*** Highly significant for p < 0.001. The parameters not shown, had no statistical significance in the analysis of data.

C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223 7219

collection of samples from the deposit of the washingmachine. The selected independent variables (period/yearof campaign, and beginning, middle or end of the cam-paign) did not affect the values for these parameters.

Relative high values of K, P, and reduced and oxidizedN forms were observed in general. The OWW collectedduring the second campaign were richer in K and P whatmay have resulted from the solubilization of these com-pounds in the waters left in deposit for some time, whilethe samples from 2003/2004 were collected immediatelyafter washing. It is obvious from Table 3 that these param-eters were negatively correlated with pH values, which con-firms our previous statement. The analysis of the C/N ratiois of particular interest since it is in general very high. Thenitrogen is mainly in the organic form and very little in theoxidized forms (Tables 2 and 3). These two facts may indi-cate low microbial activity in the communities able to min-eralise the organic carbon and the organic nitrogen,although no significant statistical correlations could befound between microbial counts and nitrogen contents.

The OCW samples were collected directly from the ver-tical centrifuge. The mean temperatures registered werehigher than those from OWW, varying from 22 to 32 �Cas a consequence of the operation that precedes centrifuga-tion. These waters were also more acidic (pHmean 5.9) as aresult of the free fatty acids dilution in the aqueous med-ium. Neither pH nor Temp had significant variationsbetween the two extraction campaigns. The mean valuefor DO was 6.0 mg O2/L, which was very low for oxidationof these waters, considering that the mean COD valueswere 24,000 mg O2/L. The BOD values were again verylow, which was reflected in COD/BOD ratios. For thesewaters, only 23.3% of the organic load could be biolo-gically degraded by aerobic processes. Mean valuesfor polyphenols in the second extraction period were883 mgcafeic acid/L, in contrast with 34.6 mgcafeic acid/Lobtained in the previous campaign. Curiously, the meanvalues for this parameter both in OWW, and in OCW werevery similar, although higher values would be expected inthe centrifuge water samples. Most probably the main frac-tion of polyphenolic compounds is more soluble in the oily

phase, and was retained by the olive oil while in centrifuga-tion. Only the more polar fraction of polyphenols was sol-uble in the aqueous medium, and that is why the values forthis parameter in OCW were very similar to those obtainedfor OWW. During the first campaign year the oil andgrease content registered the highest value in the secondsampling date, while in the second campaign the samewas observed but in the last sampling date. This parameterwas highly correlated with the values of total solids andCOD and in agreement with the variation in total solidscontent during both sampling years (Tables 2 and 4). Theseextremely high values were probably a consequence of atemporary failure in the grinding system, which led to theaccumulation of olives for a longer period of time. Themicronutrients P, K, and especially Norg were more abun-dant and positively correlated with the higher DOC,COD, total solids and oil and grease values obtained in thissampling point when compared with the ones obtained forOWW (Table 4).

The OCW and the OMW had basically the same originbut the second were subjected to deposit for long periods oftime. The mean pH values (pHmean 5.36) measured inOMW were acidic and mean temperatures were similar tothose obtained in OCW (Tmean = 25.5), but the mean DOvalues were very low (1.47 mg O2/L). The organic load(in terms of COD) increased, relatively to OWW andOCW, to mean values of 28,252 mg O2/L. These highorganic matter values, together with very low DO contentsmake these waters highly difficult to treat by biologicalmeans. The measured COD values showed fluctuationsalong time and the same was observed for other relatedparameters, as total solids and DOC values. These resultswere probably related with sedimentation phenomenaoccurring in the stabilization basin, where the samples werecollected. The oil and grease contents were highly corre-lated with the above mentioned parameters (Table 5).The nutrient contents presented a marked increase rela-tively to what was observed in OCW. At this samplingpoint the values of K and P were very high and highly cor-related, not only between each other, but also with DOC,COD, BOD, total phenols and oil and grease values. This

Table 4Pearson correlations in physicochemical parameters measured in samples from OCW, during the two campaign periods

OCW correlations N = 6 Temp COD BOD DOC Total solids Oil and grease Total phenols K P NKj Norg

Temp 1.0COD ns 1.0BOD 0.95* ns 1.0DOC ns 0.98** ns 1.0Total solids ns 1.00*** ns 0.98** 1.0Oil and grease ns 0.97** ns 0.97** 0.97** 1.0Total phenols 0.98** ns ns ns ns ns 1.0K ns 0.98** ns 1.00*** 0.98** 0.99** ns 1.0P ns 0.98** ns 0.99*** 0.98** 0.99*** ns 1.00*** 1.0NKj ns 0.98** ns 0.99** 0.98** 0.98** ns 1.00*** 1.00*** 1.0Norg ns 0.98** ns 0.99** 0.98** 0.99** ns 1.00*** 1.00*** 1.00*** 1.0

ns: Not significant.* Significant for p < 0.05.

** Very significant for p < 0.01.*** Highly significant for p < 0.001. The parameters not shown, had no statistical significance in the analysis of data.

Table 5Pearson correlations in physicochemical parameters measured in samples from OMW during the two campaign periods

OMW correlationsN = 6

pH COD BOD DOC Totalsolids

Oil andgrease

Totalphenols

K P NKj Namon Norg

pH 1.0COD 0.96** 1.0BOD ns ns 1.0DOC ns 0.90* 1.00*** 1.0Total solids 0.99** 0.98** ns ns 1.0Oil and grease 0.91* 0.97** ns 0.91* 0.96** 1.0Total phenols ns ns 0.98** 0.96** ns ns 1.0K ns 0.92* 0.99** 1.00*** ns 0.92* 0.94* 1.0P ns 0.93* 0.98** 0.99*** ns 0.92* 0.93* 1.00*** 1.0NKj 0.96** 1.00*** ns 0.88* 0.99** 0.98** ns 0.91* 0.91* 1.0Namon 0.96** 0.94* ns ns 0.98** 0.95* ns ns ns 0.97** 1.0Norg 0.96** 1.00*** ns 0.88* 0.99** 0.98** ns 0.91* 0.91* 1.00*** 0.96** 1.0

ns: Not significant.* Significant for p < 0.05.

** Very significant for p < 0.01.*** Highly significant for p < 0.001. The parameters not shown, had no statistical significance in the analysis of data.

7220 C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223

was probably due to dissolution after variable periods oftime, together with low pH values which, as seen forOWW, favour nutrient dissolution. Regarding nitrogencontents in these waters, in general the ratios C/N werevery high, and nitrogen contents (especially Norg) were alsocorrelated with parameters like COD, oil and grease, totalsolids, DOC and micronutrients (Table 5).

In order to determine which groups of physicochemicalparameters could be useful to characterize the samplingpoints and/or distinguish collection dates a principal com-ponents analysis (PCA) was performed. Parameters as pH,total phenols, oil and grease, Namon and Nnit were not ableto separate the three sampling points, while others likeTemp, COD, BOD, DOC and total solids generally pre-sented different values for OWW, distinguishing this envi-ronment from OCW and OMW. P, K, Norg and NKj

measurements were significantly different in the threesampling points. The DO content was the only parameterthat could aggregate OCW with OWW, presenting signi-

ficant different values for OMW. The best factorialplane formed by the first two extracted factorial axes (F1and F2) accounted for 69% of the total sample variance(58% for F1, 11% for F2). The PCA confirmed the correla-tions obtained, establishing three groups, which corre-sponded to the selected sampling points – OWW, OCWand OMW. Based on Kaiser’s criterion (eigenvalues >1)four factorial planes could be extracted from PCA. Table6 presents the calculated Pearson correlations betweenextracted factorials and the analysed physicochemicalparameters.

Based on these correlations, and considering the repre-sentation of the sampling points on the factor planes (datanot shown) it can be seen that samples from OWW havelow COD, BOD, and DOC, oil and grease and nutrientscontents. On the other hand, based on Factor 2 which ishighly correlated with Temp and DO parameters, it canbe concluded that samples from OMW have the highestmean temperatures and the lowest DO values.

Table 6Pearson correlations calculated for the physicochemical parameters andthe most relevant factorial planes extracted from principal componentsanalysis (PCA)

Physicochemicalparameters

Factor 1,E = 8.758

Factor 2,E = 1.604

Factor 3,E = 1.426

Factor 4,E = 1.274

DO ns �0.76** ns nsTemp ns 0.55* ns nsCOD �0.88*** ns ns nsBOD �0.78*** ns ns nsDOC �0.97*** ns ns nsTotal solids �0.72** ns ns nsOil and grease �0.87*** ns ns nsTotal phenols ns ns ns �0.71**

K �0.97*** ns ns nsP �0.97*** ns ns nsNKj �0.98*** ns ns nsNamon �0.79*** ns ns nsNnit ns ns 0.65* �0.60*

Norg �0.98*** ns ns ns

E values represent the ‘‘eigenvalues” of each factorial plane extracted. ns:Correlation not significant.

* Significant correlation for p = 0.05.** Very significant correlation for p = 0.01.

*** Highly significant correlation for p = 0.001.

C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223 7221

According to our results and to previous reports byother authors (Paredes et al., 1999), OMW present charac-teristics such as high levels of K, N, P, Ca and Mg, impor-tant factors for soil fertility, that could make these waterssuitable for soil fertilization. This could be a low-cost eco-nomic way to solve part of the environmental probleminherent to their disposal (Paredes et al., 1999, 2000,2001; Casa et al., 2003; Albuquerque et al., 2004). On theother hand, their content in fats and polyphenols usuallylimits their direct application to soils. This was demon-strated by several phytotoxicity studies, which showed thatOMW toxic effect is reflected, not only on seed germina-tion, but also on plant growth (Tsioulpas et al., 2002; Casaet al., 2003). From these results we can conclude that thecharacterization of OMW is fundamental before decidingits best use, whether it is their direct application on soils,or some other kind of physicochemical or biologicaltreatment.

3.3. Correlation between microbial counts and

physicochemical parameters

A statistical analysis was performed in order to identifypossible correlations between physicochemical parametersand microbial counts in the three sampling points. Consid-ering the normal distribution of the samples the ANOVAvariance analysis for the microbiological parametersshowed that only OWW could be discriminated in relationto the other two selected sampling points. The counts forthe three groups of microrganisms overlapped in OCWand OMW, and were significantly different than thoseobtained in OWW. The significance level was not the samefor all groups, that is, for heterotrophs the differences were

very significant (p < 0.01), while for moulds and yeasts thedifferences were significant (p < 0.05). The populations offilamentous fungi and yeast were very similar in abundanceboth in OCW and in OMW samples. Factorial analysesconsidering the counts of the three microbial groupsshowed no capacity to distinguish, neither the samplingpoints, nor the collection dates (data not shown). Never-theless it is noteworthy that, despite the increase in envi-ronmental toxicity, the heterotrophs seemed to be themost affected, since in OWW the mean counts(9.11 � 108 CFU/mL) were significantly higher than theones obtained in OCW and OMW for this group(7.15 � 106, 3.34 � 107, respectively). As for filamentousfungi and yeast groups their decrease from OWW to theother sampling points was not so marked, indicating thatthese are microorganisms better adapted to this environ-ment. In fact several authors have tested a diversity of fun-gal isolates (Aggelis et al., 2003; Tsioulpas et al., 2002;Fountoulakis et al., 2002; Fadil et al., 2003; Dhouibet al., 2006; Jaouani et al., 2003; Dias et al., 2004; Robleset al., 2000), or yeast isolates of the genera Candidae andYarrowia (Fadil et al., 2003; Martinez-Garcia et al., 2007;Papanikolaou et al., 2007; D’Annibale et al., 2006; Ettayebiet al., 2003; Lanciotti et al., 2005; Giannoutsou et al., 2004)to treat OMW.

Most of the literature available, refer that the toxicity ofOMW, besides their high organic loads, is due to their con-tent in phenolic compounds and free fatty acids (Sayadiet al., 2000; Sassi et al., 2006). White-rot fungi are amongmicroorganisms able to degrade fractions of the phenoliccompounds, namely those with low-molecular masses usingligninolytic enzymes like lignin peroxidase (LiP) or manga-nese dependent peroxidase (MnP) (Sayadi et al., 2000).However, the toxic effect in treated OMW does not disap-pear, probably due to the formation of more toxic productslike phenoxy radicals, or because ligninolytic enzymes areinhibited by high-molecular mass phenolic fractions (Say-adi et al., 2000) which continue recalcitrant after fungitreatment. Paixao et al. (1999) proved the high toxicity ofOMW, independently of the extraction procedure used.In the same study, the authors emphasized the difficultiesin determining which of the chemical parameters ofOMW were the most significant in causing toxicity. Thepresence of microcrustacean species like Tamnocephalus

platyurus and Daphnia magna, revealed high correlationswith parameters like COD, solids, nitrogen, and tanninsand lignins, but no correlations with the most referredparameters, like phenols and oil and grease. In fact, studiesusing filamentous fungi isolates (Sayadi et al., 2000) oryeasts (Peixoto et al., 2007) to pre-treat OMW showed thatthe decrease in tannin concentrations decreased the inhibi-tory effect of these waters, while others, using fungi to treatOMW revealed that the decrease in toxicity was not pro-portional to the removal of phenolic compounds (Tsioul-pas et al., 2002).

The results here presented approach these two last stud-ies, since the polyphenol content variation was not fol-

7222 C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223

lowed by correlated variations in CFU counts in none ofthe tested microbial groups, or sampling points. On theother hand, the increase in organic load mainly COD,DOC, total solids and oil and grease content, as well asthe decrease in pH and DO seemed to affect the CFUcounts. In the present work high numbers of microbialpopulations were obtained, particularly filamentous fungiand yeasts and these results point to a low OMW toxicictyeffect for these two groups of microorganisms.

Considering this, it is suggested that OWW could beused for soil application or reused for washing purposesafter simple pre-treatment as low-cost procedures to dis-pose of these wastewaters. Treatments may be performedby several processes described by many authors, such ascomposting (Paredes et al., 2002), anaerobic digestion(Dhouib et al., 2006), enzymatic treatment (Casa et al.,2003), aerobic treatments (Garrido Hoyos et al., 2002)among others. In all, it is necessary a pre-treatment processto eliminate the phenolic toxic effect which could beachieved by using microorganisms directly isolated fromthe same waters. Our group is now undertaking an identi-fication process of about two hundred yeast isolatesobtained from this survey. Preliminary results (not shown)indicate that these yeast strains are well adapted to thestressing environmental conditions, and therefore are goodcandidates for OMW pre-treatment.

4. Conclusions

Physicochemical characterization of waters generated byolive oil processing plants is crucial to decide the best wayto dispose or to reuse them. COD, DOC, K, P and N con-tents are sufficient for a rapid and less costly chemical char-acterization of these waters, and can discriminate OWWfrom the other effluents (OCW and OMW). It is shownthat besides polyphenols and fatty acids other factors likelow pH, low DO contents, high solids and organic mattercontents also contribute to OMW toxicity. Yeast isolatesfrom OMW could be used in aerobic pre-treatment pro-cesses to eliminate some of the OMW toxic effects, sincethey are well adapted to OMW toxicity.

References

Aggelis, G., Iconomou, D., Christou, M., Bokas, D., Kotzailias, S.,Christou, G., 2003. Phenolic removal in a model olive oil millwastewater using Pleurotus ostreatus in bioreactor cultures andbiological evaluation of the process. Water Research 37, 3897–3904.

Albuquerque, J.A., Gonzalvez, J., Garcıa, D., Cegarra, J., 2004. Agro-chemical characterization of ‘‘alperujo”, a solid by-product of the two-phase centrifugation method for olive oil extraction. BioresourceTechnology 91, 195–200.

Aktas, E., Imre, S., Ersoy, L., 2001. Characterization and lime treatmentof olive mill wastewater. Water Research 35 (9), 2336–2340.

APHA, 1992. Standard Methods for the Examination of Waters andWastewaters, 18th ed. American Public Health Association, Washing-ton, DC.

Casa, R., D’Annibale, A., Pieruccetti, F., Stazi, S.R., Sermanni, G.G., LoCascio, B., 2003. Reduction of the phenolic components in olive-mill

wastewater by an enzymatic treatment and its impact on durum wheat(Triticum durum Desf.) germinability. Chemosphere 50, 959–966.

D’Annibale, A., Ricci, M., Quaratino, D., Federici, F., Fenice, M., 2004.Panus tigrinus efficiently removes phenols, color and organic load fromolive-mill wastewater. Research in Microbiology 155, 596–603.

D’Annibale, A., Sermanni, G.G., Federici, F., Petruccioli, M., 2006.Olive-mill wastewaters: a promising substrate for microbial lipaseproduction. Bioresource Technology 97 (15), 1828–1833.

Dias, A.A., Bezerra, R.M., Nazare-Pereira, A., 2004. Activity and elutionprofile of laccase during biological decolorization and dephenolizationof olive mill wastewater. Bioresource Technology 92, 7–13.

Dhouib, A., Aloui, F., Hamad, N., Sayadi, S., 2006. Pilot-plant treatmentof olive mill wastewaters by Phanerochaete chrysosporium coupled toanaerobic digestion and ultrafiltration. Process Biochemistry 41, 159–167.

Ettayebi, K., Erradichi, F., Jamai, L., Tahri-Jouti, M.A., Sendide, K.,Ettayebi, M., 2003. Biodegradation of polyphenols with immobilizedCandida tropicalis under metabolic induction. FEMS MicrobiologyLetters 223, 215–219.

Fadil, K., Chahlaoui, A., Ouahib, A., Zaid, A., Broja, R., 2003. Aerobicbiodegradation and detoxification of wastewaters from the olive oilindustry. International Biodeterioration and Biodegradation 51, 37–41.

Fiorentino, A., Gentili, A., Isidori, M., Monaco, P., Nardelli, A., Parrella,A., Temussi, F., 2003. Environmental effects caused by olive Millwastewaters: toxicity comparison of low-molecular-weight phenolcomponents. Journal of Agricultural and Food Chemistry 51, 1005–1009.

Fountoulakis, M.S., Dokianakis, S.N., Kornaros, M.E., Aggelis, G.G.,Lyberatos, G., 2002. Removal of phenolics in olive mill wastewatersusing the white-rot fungus Pleurotus ostreatus. Water Research 36,4735–4744.

Giannoutsou, E.P., Meintanis, C., Karagouni, A.D., 2004. Identificationof yeast strains isolated from a two-phase decanter system olive oilwaste and investigation of their ability for its fermentation. Biore-source Technology 93, 301–306.

Garrido Hoyos, S.E., Martinez Nieto, L., Camacho Rubio, F., RamosCormenzana, A., 2002. Kinetics of aerobic treatment of olive-millwastewater (OMW) with Aspergillus terreus. Process Biochemistry 37,1169–1176.

Jaouani, A., Sayadi, S., Vanthournhout, M., Penninckx, M.J., 2003.Potent fungi decolourisation of olive mill wastewaters. Enzyme andMicrobial Technology 33, 802–809.

Lanciotti, R., Gianotti, A., Baldi, D., Angrisani, R., Suzzi, G., Mastro-cola, D., Guerzoni, M.E., 2005. Use of Yarrowia lipolytica strains forthe treatment of olive mill wastewaters. Bioresource Technology 96,317–322.

Martinez-Garcia, G., Johnson, A.C., Bachman, R.T., Williams, C.J.,Burgoyne, A., Edyvean, R.G.J., 2007. Two-stage biological treatmentof olive mill wastewater with whey as co-substrate. InternationalBiodeterioration and Biodegradation 59, 273–282.

Mouncif, M., Tamoh, S., Faid, M., Achkari-Begdouri, A., 1993. A studyof chemical and microbiological characteristics of olive mill wastewater in Morocco. Grasas e Aceites 44 (6), 335–338.

Paixao, S.M., Mendonc�a, E., Picado, A., Anselmo, A.M., 1999. Acutetoxicity evaluation of olive mill wastewaters: a comparative study ofthree aquatic organisms. Environmental Toxicology 14, 263–269.

Papanikolaou, S., Galiotou-Panayotou, M., Fakas, S., Komaitis, M.,Aggelis, G., 2007. Citric acid production by Yarrowia lipolytica

cultivated on olive-mill wastewater-based media. Bioresource Tech-nology. doi:10.1016/j.biortech.2007.05.005.

Paredes, C., Cegarra, J., Roig, A., Sanchez-Monedero, M.A., Bernal,M.P., 1999. Characterization of olive mill wastewater (alpechin) and itssludge for agricultural purposes. Bioresource Technology 67, 111–115.

Paredes, C., Roig, A., Bernal, M.P., Sanchez-Monedero, M.A., Cegarra,J., 2000. Evolution of organic matter and nitrogen during co-composting of olive mill wastewater with solid organic wastes. BiologyFertility of Soils 32, 222–227.

C. Amaral et al. / Bioresource Technology 99 (2008) 7215–7223 7223

Paredes, C., Bernal, M.P., Roig, A., Cegarra, J., 2001. Effects of olive millwastewater addition in composting of agroindustrial and urban wastes.Biodegradation 12, 225–234.

Paredes, C., Bernal, M.P., Cegarra, J., Roig, A., 2002. Bio-degradation ofolive mill wastewater sludge by its co-composting with agriculturalwastes. Bioresource Technology 85, 1–8.

Peixoto, F., Martins, F., Amaral, C., Gomes-Laranjo, J., Almeida, J.,Palmeira, C.A., 2007. Evaluation of olive oil mill wastewater toxicityon the mitochondrial bioenergetics after treatment with Candida

oleophila. Ecotoxicology and Environmental Safety. doi:10.1016/j.ecoenv.2007.11.003.

Peres, J.A.S., 2001. Degradac�ao de acidos fenolicos por processos deoxidac�ao avanc�ados e pre-tratamento de aguas ruc�as pelo reagente deFenton. PhD Thesis. UTAD, Vila Real, Portugal.

Robles, A., Lucas, R., de Cienfuegos, G.A., Galvez, A., 2000. Biomassproduction and detoxification of wastewaters from the olive oil

industry by strains of Penicillium isolated from wastewater disposalponds. Bioresource Technology 74, 217–221.

Sassi, A.B., Boularbah, A., Jaouad, A., Walker, G., Boussaid, A., 2006. Acomparison of Olive oil Mill Wastewaters (OMW) from three differentprocesses in Morocco. Process Biochemistry 41 (1), 74–78.

Sayadi, S., Allouche, N., Jaoua, M., Aloui, F., 2000. Detrimental effects ofhigh molecular-mass polyphenols on olive mill wastewater biotreat-ment. Process Biochemistry 35, 725–735.

Tsioulpas, A., Dimou, D., Iconomou, D., Aggelis, G., 2002. Phenolicremoval in olive oil mill wastewater by strains of Pleurotus spp. inrespect to their phenol oxidase (laccase) activity. Bioresource Tech-nology 84, 251–257.

Vitolo, S., Petarca, L., Bresci, B., 1999. Treatment of olive oil industrywastes. Bioresource Technology 67, 129–137.