Fingerprints of some chlorinated hydrocarbons in plant foliage from Africa

Fingerprints of some chlorinated hydrocarbons in plant foliage from Africa

Chemosphere,Vol.27,No.l 1, pp 2235-2252, 1993 Printed in Great Britain 0045.6535/93 $6.00 + 0.00 Pergmnon PressLtd. Fingerprints of Some Chlorinated...

882KB Sizes 1 Downloads 18 Views

Chemosphere,Vol.27,No.l 1, pp 2235-2252, 1993 Printed in Great Britain

0045.6535/93 $6.00 + 0.00 Pergmnon PressLtd.

Fingerprints of Some Chlorinated Hydrocarbons in Plant Foliage from Africa

P. Tremolada 1, D. Calamari 1., C. Gaggi2 and E. Bacci2. l Groupof Ecotoxicology Institute of AgriculturalEntomology,Universityof Milan. Via Celoria2, 1-20133 Milan, Italy 2Departmentof EnvironmentalBiology,Universityof Siena. Via delle Cerchia 3, 1-53100 Siena, Italy (Received in Germany 11 May 1993; accepted 26 July 1993)

ABSTRACT Concentrations ofDDTs, HCHs and HCB in foliage have been used to evaluate the contamination levels of two African areas and Seychelles and Mauritius Islands. Physico-chemical properties in combination with environmental features play the most important role in the global distribution of chlorinated hydrocarbons. However past and present applications appear to be significant in determining the characteristics of the contamination pattern in different areas. Results are discussed in comparison to previous African data. Remarkable differences in contamination levels are evident among the areas considered. Log-Probit and Correspondence Factor Analyses are used for the characterization of the typical distribution pattern of each area. Relative differences in the composition of the pollutant mixture (HCHs dominance, different DDE/DDT ratios etc.) seem to indicate a "fingerprint" of the contamination for each geographical-economical homogeneous region.

INTRODUCTION Chlorinated hydrocarbons, such as DDT and metabolites, HCH isomers and HCB have been detected in all the continents; they have reached the most remote areas by a long-range transport process due to the atmosphere where they are distributed mainly in the vapour phase (Risebrough et al., 1976). Vegetation has been used as a sampling tool for the evaluation of the tropospheric contamination level of an area (Calberg et al, 1983; Gaggi et al., 1985). Vapours of these substances in the air are able to be absorbed by plant leaves, mosses and lichens and constitute the main way of contaminating vegetation (Nash and Beall, 1970; Bacci and Gaggi, 1986; Bacci et al., 1990; Trapp et al., 1990). When the exposure time is long enough (i.e.: after the life cycle of a leaf), leaf/air equilibrium is approached and concentration in plant leaves may be

2235

2236

used to calculate the concentration in air by means of the bioconcentration factor according to the "azalea model" (Bacci et al., 1990; Paterson et al., 1991). In order to describe the global contamination levels and the main distribution trends, a north-south distribution pattern in the foliage for HCHs, HCB and DDTs compounds was proposed by Calamari et al. (1991). Tropical regions have been indicated as the main source areas of these compounds while polar and subpolar regions, as well as other cold remote sites (i.e. high mountain tops), where low temperatures enhance deposition phenomena (Calamari et al., 1991), behave as receivers for the above mentioned global contaminants. Environmental features, such as the air temperature, in combination with physieo-ehemical properties have been indicated as the most important factors in the distribution pattern of remote areas. Another significant factor determining the distribution of organochlorine insecticides deals with the observation that DDTs in tropical regions and HCHs in the northern hemisphere have been revealed as the most abundant contaminants, in accordance with their actual or recent major use. The aim of this work is to better understand the distribution patterns, the contamination intensity and possible environmental processes of these chemicals in the African continent. In Sierra Leone, Lake Victoria, Seychelles and Mauritius Islands, four new transects of mango leaf samples were analyzed for chlorinated pesticides. Results have been compared with previously published similar findings from nine other African transects and findings from Tristan da Cunha Islands (South Atlantic) (Bacci et al., 1988; Calamari et al., 1991). The last analytical results and the previous ones, are discussed in relation to the hypothesis of the typical distribution pattern of an area as a "fingerprint" of the past and present use and for the comprehension of the main environmental processes that could have occurred.

MATERIALS AND METHODS Sample collection Foliage samples of mango leaves (Man~feraindica) were collected in two areas in the African continent (Sierra Leone, Lake Victoria) and Seychelles and Mauritius Islands during 1991. Tab. 1 reports the four transect sites with the locations between which the samples were collected, together with the mean latitude and lon~tude of each transect.

Tab. 1 - The four transects sites (Samplingarea), the locationsbetweenwhich the sampleswere collected(Location),the numberof samples(n°) and the mean latitudeand longitudeof each transect(Latitudeand Longitude).

Samplingarea

Location



Latitude

Longitude

Sierra Leone Lake Victoria Seychelles Mauritius

Port Loko-Kenema Kisumu-Mbita Point Victoria City Island Albion-Fond du Sal

8

6 10 6

8-9ON 0o 5os 20 ° 30' S

12°O 34°E 56°E 57 ° 30' E

2237

Fig. 1 shows the transect sites of the four transects of the present work and the sites of ten other transects, the results of which are published in Calamari et al. (1991). Transect-sampling mode was performed in order to have a sufficient representation of the area with a limited number o f samples. A variable number of samples was collected in a range from tens to hundreds of kilometres. Mango leaf samples (about 10 g each) were collected from the ground at the end of their natural life cycle and wrapped in aluminium foil, kept cold (4 °C) whenever possible, and then stored at -20 °C until analysis was performed.

Mall-Guinea ,; LBenin-Burkina Faso

I

I7

,: /

L--.. ',..

i

'

"'-¢J

(

'

Nairobi Kenya

,

•,

~';

~' , " - r ~ ,

,;~

/ t

'i

Suhum,~i"" ---'¢, l~ Ivory Cc

,--J .,

i GhanaAccra

""

,/ , ',..--., " • '

l

"

! , t ~

~-Tri~tan da Cunha

(m}

Hount

Kenya

i

..i "'i-.".....

, ;', ! (

r

i

Kenya

i~ \i,~ -~. ( ". . . . . ,~,

.(:

-

"3

~

t

eychelles

uritiu~s°

Lake Victoria

Town

Fig. 1 - Map of Africa with the indication of the transect sites. The four transects of the present work are identified by an -R-and the transects from the literature (Calamari et al. 1991) are identified by a I .

2238

Chemicals HCB, ct-HCI-I, y-HCH, o,p'DDE, p,p'DDE, o,p'DDD, p,p'DDD, o,p'DDT and p,p'-DDT were purchased from Supelco Inc., Bellefonte, PA USA. All compounds were analytical standards >99% purity, nHexane for residue analysis was purchased from giedel-de Haen, Seelze, Germany; Florisil for residue analysis 0.150-0.250 mm (60-100 mesh ASTM) and sulphuric acid 95-97% p.a. from Merck, Darmstadt, Germany and cellulose extraction thimbles from Schleicher & Schuell, Dassel, Germany.

Chemical analysis Sample preparation: after a partial oven-drying (30 °C, 12 h), samples were minced and homogenized. Residual water was measured on sub-samples (5 h at 105 °C). Extraction and clean-up: cellulose extraction thimbles were oven dried (130 °C, 4 h) and decontaminated by means of a pre-extraction in Soxhlet apparatus with n-Hexane. Samples were weighed in the extraction thimbles and the extraction was performed with Soxhlet apparatus using n-Hexane (8 h). Sulphuric acid clean-up (10 ml added to the extracts; 12 h), was operated. Florisil column chromatography (1,5 g of Florisil in 8 mm I.D. glass columns) was performed after the reduction of the volume by a rotary evaporator operated at 45 °C. Florisil was previously dried (130 °C; 4 h) and washed with n-Hexane. Samples sorbed on Florisil, were eluted with 45 ml ofn-Hexane. The volume of the samples was reduced by a rotary evaporator to 1 or 0.5 ml for the gas-chromatographic analysis. Packed column GLC-ECD: samples were analyzed by a Perkin-Elmer Sigma-3B gas chromatograph with a split-splitless injector using a borosilicate glass-columns 2 m length and with 2 mm I.D., packed with GP 4% SE-30 6% SP-2401 and GP 1.5% SP-2250 1.95% SP-2401 on 100/120 mesh Supelcoport. The carrier gas was argon-methane, 95/5%; flows were: 60 and 40 (scavenger) ml/min; the injector oven and detector temperature were 210, 200 and 280 °C, respectively. Detection limits were as follows: 0.01 ng/g dry weight for HCB, or- and y-HCH and 0. I ng/g dry weight for o,p" andp,p'DDE, DDD, DDT.

Statistical troatment of the data A Log-Probit analysis was chosen to define mean contamination level of each transect, taking into consideration samples below detection limits, after grouping them at the detection limit value (indicated above). Log-Probit analysis has already been used to elaborate this type of data (Calamari et al., 1991). The Log-Probit analysis was performed according to a BASIC computer program suggested by Trevors (1986), slightly modified. A median concentration C50 was calculated for each transect. The values corresponding to probit 4 and 6 (C16 and C84 ) were also calculated, indicating the range around the median where 68% of the results were expected. A spread parameter for C50, the "slope" S, was calculated in accordance with the formula [(C84/C50)+(C50/C 16)]/2. A ~2 test gives an indication of the homogeneity of the sample population.

2239

C50 values were utilized for the graphical representation of the typical contamination pattern of an area by normalized bar diagrams; a-HCH, HCB, T-HCH, p,p'-DDE, o,p'-DDT and p,pLDDT contamination levels in mango leaves were chosen for the bar diagram display. Mass units of contaminants in concentration data were transformed from nanograms (ng) to picomoles (pmol) for the Correspondence Factor Analysis (CFA). The CFA approach was chosen in order to give preferential emphasis to the relative differences in the compound ratios rather than the absolute values that might be misleading in the comprehension of the typical distribution pattern of an area. The analysis was performed according to a statistical computer program STAT-ITCF version 3.0, 1987 originally produced by ECOSOFT, translated and corrected by the "Institut Technique des C~r6ales et des Fourrages", Paris. The CFA is a multivariate approach showing the possible correspondence existing between observation and the attributes or variables. Observations and variables are included in a data matrix and are analyzed as vectors; the vectors are displayed as points in a dual low-dimensional vector space. The space is composed of a system of orthogonal axes', the number of two-dimansional views depends on how many combination of axes can be Obtained. The axes are linear combinations of the original variables. The dual representation of individuals and variables allows a very concise graphical display expressing a number of different features of the data in a single picture. The display shows similarities and dispersions as distances in the planar projection of the points and indicates correspondence inside the matrix (Devillers and Karcher, 1990). The graphs are a deformed representation of reality because the points are projected on the plane. Two near points or individuals on the graph might not be so in the space.

RESULTS AND DISCUSSION

Characteristics of the analvzed compounds The technical product of the hexachlorocyclexane (HCH) insecticide is a mixture of isomers: ex-HCH (about 80%) and "/-HCH (about 10%) but only the last has insecticidal properties. In several countries, a product containing the pure T-HCH is now in use, replacing the previously mentioned mixture. The ~ , ratio is considered a good indicator of the origin of the contaminants and the age of the contamination (Pacyna and Oehme, 1988). In the more technological countries, the product containing pure ~,-HCH is now used, while old technologies produce the mixture of isomers. Hexachlorobenzene (HCB) was used as a fungicide and it is a by- or co-product of many chemical syntheses; high levels of HCB are an indication of a use area or a highly industrialized area. The DDT degradation products (mainly DDE and DDD) are indicators of an old use of this insecticide or that it has been transported from a long distance; during the transfer the original DDT may be transformed in its more stable degradation products. DDT, as other low volatile persistent compounds, has been indicated as subject to a long-range transport process by a continuous cycle between soil, air and water with the so-called "gas-chromatographic effect" (Risebrough, 1990) or the "grasshoppers effect" (Calamari et al., 1991).

2240

C o n t a m i n a t i o n levels in m a n g o leaves

Concentrations ofct-HCH,

HCB,

),-HCH, o,pLDDE, p , p ' - D D E ,

o,p'DDD,

p,p'DDD,

o,p'DDT,

p . p ' D D T in m a n g o leaves in t h e different African states are reported in Tab. 2.

Tab. 2 - Concentrations in ng/g dry weight of ct-HCI-I, HCB, ?-HCH, o,p'DDE, p,p'DDE, o,p'-DDD, p.p'-DDD, o,p'DDT and mango leaf samples in Africa. Samples arc identified by a transect location code (SL = Sierra Leone, LV = Lake Victoria, SE = Seychelles and MA -- Mauritius) and are divided in transects.

p,p'-DDTin

ng/g d.w. Sample SL-I SL-2 SL-3 SL-4 SL-5 SL-6 SL-7 SL-8 LV-1 LV-2 LV-3 LV-4 LV-5 LV-6 SE-I SE-2 SE-3 SE-4 SE-5 SE-6 SE-7 SE-8 SE-9 SE-10 MA-I MA-2 MA-3 MA-4 MA-5 MA-6

ct-HCH

HCB

y-HCH

o,p'DDE

p,p'DDE

o,p'DDD

p,p'-DDD

o,p'-DDT

p,p'DDT

0.06 0.13 0.19 4.4 0.08 0.14 0.1 0.1 0.24 0.18 0.10 0.13 0.10 0.24 1.4 1.5 1.2 14 1.6 8.0 19 0.58 1.7 4.9 0.16 0.27 0.58 0.25 0.21 0.20

ND ND ND 0.09 0.04 0.02 0.03 0.02 ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND ND

0.02 0.03 0.03 0.09 0.02 0.04 0.07 0.15 0.10 0.11 0.04 0.13 0.62 0.10 0.04 0.02 0.09 0.40 0.04 0.20 0.08 0.08 0.08 0.04 0.10 0.24 0.50 0.27 0.28 0.26

0.6 0.2 0.4 0.1 0. ! 0.1 0.1 0.1 ND 0.3 0.2 0.1 0.2 ND ND ND ND ND ND ND ND ND ND ND 0.9 3.0 11 15 3.5 3.1

1.8 0.7 1.5 0.6 0.4 0.7 0.3 0.6 1.5 4.1 4.0 0.7 3.4 0.6 ND 0.1 3.3 0.-t 1.0 0.1 ND 0.1 0.4 0.I 10 26 67 140 38 42

1.2 0.4 0.7 0.2 0.1 0,3 0.1 0.5 ND ND ND ND ND ND ND ND ND ND ND bid ND ND ND ND 2.4 16 53 28 15 2.4

13 3.0 6.1 1.4 0.7 2.1 0.7 3.0 ND 0.2 ND ND ND ND ND ND ND 0.3 ND ND ND ND ND ND 48 76 240 570 130 37

2.5 0.9 1.8 0.9 0.4 0.8 0.3 0.7 0.2 0.9 0.2 0.3 0.4 ND ND 0.5 1.1 0.4 0.6 ND ND ND 0.3 hid 4.9 39 120 87 28 17

24 6.4 14 5.1 19 4.7 1.8 6.5 1.4 36 1.1 10 1.9 1.3 N'D N'D 7.8 0.3 77 0.9 0.7 0.6 3.6 0.7 81 170 530 1300 250 200

ND= below detection limits: 0.01 ng/g dry weight for HCB, ¢t- and y - H C H and 0.1

p,p'DDE,

ng/g dry weight for

o,p'- and

DDD, DDT.

By the analysis o f the data o f the Tab. 2, a low H C H and H C B

presence and high D D T contamination

appear as the main feature o f pollution status o f these areas. H C H levels in all the transects are generally low in terms o f the present global b a c k g r o u n d , with s o m e s p o t s o f ct-HCH mainly in the Seychelles. T h e H C B levels are below the detection limits (0.01 ng/g d.w.), with a few exceptions in Sierra L e o n e s a m p l e s where, in any

2241

case, they never reach 0.1 ng/g d.w.. For the distribution of DDT and related compounds, among the different transects, major differences in the concentration levels of these pollutants can be noticed: the Mauritius samples indicate a typical hot-spot condition, with high concentrations of all DDTs, while those of the Seychelles show low concentration values of both DDT and me~abolites. However a generally high contamination status of the DDT compounds in the areas considered is evident. High DDT and low HCH and HCB levels ones seem to be the typical distribution pattern within the African continent, probably deriving from recent and abundant DDT use and from the absence of very intense sources of the more technological products, such as pure 7-HCH insecticide or the industrial contaminant HCB. The same pattern was indicated by Calamari et al. (1991): tropical countries, such as Africa, are generally characterized by a high-DDT and metabolite contamination level in the foliage, by lower levels of the more technological products, such as pure T-HCH insecticide, and by very low levels of HCB (<0.1 ng/g d.w.). Within this general feature of a high DDT contamination of the African continent and low HCH and HCB levels, each area can be considered separately. In fact, as it is important to define the mean contamination level of vast areas it is also important to clarify the main differences existing among countries. In areas of utilization of the above mentioned molecules, the history of the use of the different chemical products could affect the distribution patterns more than physico-chemical properties or environmental characteristics. If the history of use is the main factor affecting the contamination levels in use areas, even zones that present the same geographical and climatic characteristics may have different distribution patterns. By this hypothesis, each area shows a typical level and composition of the contaminants considered, called the "fingerprint " of the area. By the "fingerprint" analysis of a homogeneous area, the past and present use of these chemicals can be deduced and so the technological level of agriculture and the socio-economical conditions can be inferred.

Correspondence Factor Analysis of the contamination levels of the single samples Data can be analyzed with Correspondence Factor Analysis (CFA) and the simultaneous representation of both samples and variables allows the uniformity or the differences between the samples (distances between them) and the association with the more representative compounds (distances between samples and variables) to be recognized. The advantage of the single sample analysis is the possibility of confirming the degree of similarity of the samples coming from the same transect and the true relative differences with the other samples without any reduction of variability. In the graph the samples are associated in clouds, allowing a synthetic treatment of the data, such as the geometric mean or the Log-Probit analysis. The graphical representation of the thirty African samples from Sierra Leone, Lake Victoria, Seychelles and Mauritius is shown in Fig. 2.

2242

p,/o'- DDT SL-1 SL-2 SL-3 MA-I

rMA-3I I

IMA-,tt I MA~..~qp~DDD SL-4 5E-S

SE-2,

$E-9

SE-IO SE-7 Q.-HCH SEs-~6 SE'I

sE-3

NC8 SE-8 LV-2 LV-1 LV-6 LV-3 LV-4

LV-5 }'-HCH

Fig. 2 - Simultaneousrepresentationof the nine chemicals analyzed (ct-HCH, HCB, y-HCH, o.p'-DDE, p,p'-DDE, o.p'-DDD, p,p'-DDD, o,p'-DDTandp,p'-DDT) and of the thirty samples coming from Sierra Leone (SL), Lake Victoria (LV), Seychelles(SE) and Mauritius (MA); the samplesare identifiedby the same codes as in Tab. 2. The plane of the graphical representationof the CFA analysis is determined by the first axis in horizontal and the second axis in vertical. * Group of samples and variables laying upon the o,p'-DDDvariable. ** Two samplesand one variable laying upon the o.p'-DDEvariable.

The distribution of the samples is quite homogeneous and results, associated principally with the DDT variables, lay in a very concentrated cloud upon the intersection of the axes. Only two groups of samples are unrelated to DDT: samples from Seychelles and from Lake Victoria. The first group seems to be strongly characterized by c~-HCH compound, most of the samples lie in a cloud together with c~-HCH variable at the end of axe 1. The second group is composed by the samples from Lake Victoria, they appear to be characterized by 7-HCH variable, lying in a cloud upon the second axis. All the other samples of Sierra Leone and Mauritius are very closely associated, underlying their similarity in the relative composition of the compounds. The Seychelles samples seem to be composed of a double population in relation to the contamination patterns: one characterized by the highest o~-HCH contamination and the second by a relatively higher one for DDT with the same high c~-HCH level, so the Seychelles transect can be assumed to be a ~-HCH characteristic transect with some relatively high DDT contamination samples probably coming from some locally contaminated areas. The other transects show a homogeneous and representative contamination pattern of the origin region, indicating that each area can be characterized by a typical contamination pattern of chlorinated hydrocarbons

2243

Bar diagr~m¢ of the contamination levels in mango leaves in Africa Samples coming from the same area and having the same contamination pattern can be considered as a homogeneous population with its own internal variability, indicating more scattered or more uniform contamination sources, and a median contamination level. The geometric mean or a Log-Probit analysis is a useful system to obtain a representative value of the contamination level of an area. Tab. 3 reports the results of a Log-Probit analysis of the samples coming from the four new African transects. Six compounds were chosen from the nine analyzed to characterize the "fingerprint" of an area because of the major effectiveness of the selected compounds in discriminating the different African patterns.

Tab. 3 - Median concentration (C50 in ng/g Dry Weight) and Statistical Parameter (X2), Degrees of Freedom (DF) and spread parameter ($) of the Log-Probit Line for ct-HCH, HCB, 7-HCH, p,p'.DDE,o,p'.DDT,p,p'-DDTin the mango leaf samples in Africa.

~-HCH

HCB

y-HCH

p,p'-DDE

o,p'-DDT

p,p'-DDT

C50 S DF X2

0,11 1.6 5 0.38

0,02 2.6 3 0.57

0.04 5 1.2

0.6 2.0 5 1.6

0.7 2.2 5 1.1

4.7 2.6 5 1.4

C50

ND

DF X2

0.14 1.7 3 0.56

0.09 2.0 3 2.0

1.6 3.4 3 0.53

0.2 2.0 3 0.49

1.3 1.4 3 0.36

C50 S DF X2

2.4

ND

3.7 7 3.1

0.06 2.3 7 2.6

0.1 5.2 5 2.3

0.1 5.0 3 1.2

0.6 7.2 6 1.4

ND

0.22 2.0 3 2.5

31 2.6 3 0.61

24 4.2 3 0.27

210 2.5 3 0.38

Transect

Sierra Leone 2.2

Lake Victoria

S Seychelles

Mauritius C50 S DF

X2

0.22 1.3 3 0.25

ND= below detection limits.

The bar diagrams of the C50 values of the chlorinated hydrocarbons in the four transects are reported in Fig. 3.

2244

r~g d.v~

/

Z__

y

t4|

3"

2~ 2. Is.

ae4

L

//

0--

1

o// /

.

Lake Victoria

Sierra Leone

1(o

,

Mauritius

tl,

°'oV1t Seychelles

Fig. 3 - Bar diagramsof the median concentration(C50) in mango leavesof
Sierra Leone and Mauritius are almost identical and present the typical pattern of the African continent with a high DDT contamination and very low HCH and HCB levels. The DDT/DDE ratio is high showing ap,p" DDT dominance. Direct DDT applications are evident in both countries, while HCH and HCB show a background contamination level. The DDT composition suggests recent and continuous treatment of this insecticide on a large scale. In Mauritius the DDT contamination intensity is a result of heavy and concentrated use in the monitored areas. Lake Victoria has the same DDT dominance but the DDT/DDE ratio indicates an old contamination status with a higher presence of DDE. The contamination intensity is very low and homogeneous, suggesting an indirect use for both DDT and HCH; they probably come from the surrounding, more contaminated regions. The levels of DDT and y-HCH could be an indication of the proximity to the origin sources. The Seychelles show a very different spectrum, dominated by high a-HCH level in relation to very low yHCH and DDT concentration. The Seychelles spectrum is similar to remote area patterns, mostly characterized by contamination coming from long-range transport processes. The relatively high DDT level indicates perhaps a light, direct use which can also be deduced by the high variability of the single data that present some scattered high contaminated samples (secondary population). The results proposed here can be easily compared and supported with ten other African transects of mango leaf samples, as published by Calamari et al. (1991). Tab. 4 reports the C50 values and the slopes of the ten transect data, as published by Calamari et al. (1991).

2245

Tab. 4 - Median concentration (C50) in ng/g d.w. and spread parameter, between btafkets, of the Log-Probit Line for chlorinated hydrocatbuns in 10 transects of foliage samples collected in Africa as published by Calamari et al. (1991).

ng/g d.w. Transect

p,p'-DDE o,p'.-DDT

p,p'-DDT

a-HCH

HCB

g-Hell

Mali-Cadnca

0.50 (2.2)

<0.1

0.20 (2.2)

8. l (4.8)

4.3 (8.2)

37 (6.2)

Benin-Burkina

0.60 (2.3)

<0.1

<0.1

1.0 (11)

1.0 (5.3)

5.1 (4.7)

Ghana S u h u m

0.30 (1.7)

<0.I

2.8 (2.7)

0.40 (1.9)

0.80 (2.0)

2.4 (2.1)

Ivory Coast

0.69 (1.6)

<0.I

0.35 (1.9)

1.6 (4.9)

0.70 (1.4)

3.3 (1.8)

Ghana Accra

0.30 (1.6)

<0.I

1.0 (1.7)

1.9 (2.3)

1.5 (3.1)

15 (2.9)

Mount Kenya

7.9 (2.4)

0.52 (2.0)

0.78 (4.4)

1.1 (2.1)

1.0 (4.6)

4.0 (2.7)

Nairobi

1.4.

<0.1

0.88*

6.7*

2.1"

14"

Mombasa

2.7 (2.4)

<0.1

0.78 (1.3)

25 (1.7)

5.7 (1.6)

14 (2.O)

0.58*

0.12"

0.77*

0.60*

<0. I

4.4*

0.19 (1.5)

<0.1

<0.1

ND

ND

ND

Cape town Tristan da Cunlm

* values reported as geometric mean if samples were too few for the Log-Probit analysis. biD= below detection limits.

The bar diagrams o f the C50 values o f chlorinated hydrocarbons in the ten previous transects are shown in Fig. 4.

r~/g d.w

r,g/g d.w.

Ii

-II ,°tJ

I1'

2"1 I

151

i I

L

i;

'01 C~

a.HCH

HOB

~

p,~,.l~E O~-DDT p~.DOT

Mali-Guinea

O~/'l V e~---f/ I I." I I/ ./ ," I __L<~. a-HCH HCB O-HCH p,p'43OE o,ff-COT ptYq~OT

Cape T o w n

2246

ng/g d.w.

ng/g d.w,

16'

/_.._

14 12 10

0

e-No-~

HCB

Benin-BurkinaFaso

~

p,p'-OOE o~'43DT p,p'-OOT

Ghana Accra ng/g d.w.

ng/g d.w. 3.6 / 3

12

2,5

10

2

A

1.5 1 OS

r ~

g.M04

p,~.OD[ o,I~-DDT p,l~.Ol~

Nairobi Kenya

Ivory Coast ng/g d.w.

ng/g d.w.

,it

I

I J r

05"

~

_

.

.

, ~

Ghana Suhum

pff-[1)T

Mombasa Kenya

ng/g d.w

ro'g d.w...

0.2 / / ~ - - - - ~ 018 016 014

'l

:

0.12 0.1 008 006 004 O~

0/

g~-ICH pp'-OOE O,l~OT

HOB

g,.HC~ op'-!~E

O,d-DDT p,p',DOT

Tristan da C u n h a

oL4

~

~

~

p,p'-tX~ o.p'-l:~ p,p'-Dffr

Mount Kenya

Fig. 4 - Bar diagrams of the median concentration (C50) in mango leaves of ct-HCH, HCB. y-HCH, p,p'DDE, o.p'-DDT,p.p'-DDT for nine African transects and for Tristan da Cunha Islands, the median concentration data were published by Calamari et al (1991).

2247

Even though great differences appear from the typical contamination pattern of so many different African regions, as shown in the four previous bar diagrams, the typical African spectrum of high DDT contamination and low HCH and HCB levels is evident. Mali-Guinea, Benin-Burkina Faso and Ghana Accra reveal the same distribution pattern as Sierra Leone and Mauritius, highly dominated by p,p'-DDT level, while Ivory Coast, Nairobi Kenya and Cape Town show once more the typical p,p'-DDT dominance but with a relatively higher HCH contamination and a more abundant DDE at about half of the DDT value. A not very high contamination intensity of the last three transects is also present, indicating a more limited use of the DDT insecticide or an older contamination with an important role of transformation and transport processes. Mombasa Kenya has the same "fingerprint" as Lake Victoria with higher DDE than DDT levels, showing an old DDT contamination with a probably recent limited contribution ofp,p'DDT. Ghana Suhum has a spectrum composed of a high T-HCH level and shows a low DDT contamination if compared with the Ghana Accra transect. The high T-HCH levels indicate a local use of the pure ), isomers as insecticide (cocoa cultivation area), like the more advanced European agriculture (Calamari et al, 1993), and the DDT level, lower than in Ghana Accra, is probably an index of a neighbouring DDT use or of a very limited use. A possible origin is a short-range transport from highly-contaminated surrounding regions. The transport origin would also be indicated by a higher o,p'-DDT/p,p'-DDT ratio, because of the higher o,p' isomer volatility, (Henry's constant 10 times higher than p,p' isomer) and so higher probability of transfer during short-range transport. Mount Kenya and Tristan da Cunha bar diagrams are very different from the typical contamination pattern of the other African countries; they appear very similar to the Seychelles transect by the ct-HCH dominance among the different chlorinated hydrocarbons. As observed for the Seychelles contamination, these areas can be considered remote places where indirect or only limited contamination exists and where shortrange transport is limited in comparison with the relative importance of the global cycling of these molecules and the long-range transport processes. The intensity of the contamination is generally low, especially for the samples coming from Tristan da Cunha. The contamination of Mount Kenya samples are probably the result of many phenomena: the general contamination level is much higher in comparison with more remote areas, probably due to the proximity of very contaminated regions and to the low temperature effects as pointed out by Calamari et al. (1991). In fact high levels of HCB were detected in comparison to the absence of detectable concentrations of this compound at sea level in the surrounding regions. The relatively high DDT levels also seem to derive from the neighbouring areas. In fact, the DDT spectrum is like that of Nairobi-Kenya and the same relatively higher o,p'-DDT level as for Ghana Suhum transect, can be observed. The comparison between the contamination patterns of many related areas has proved to be very useful for establishing a general map of use intensities and use ages and for developing a possible explanation of the experimental observations on global cycling of these contaminants.

(~orrespondence Factor Analysis of the transect contamination levels The CFA is a useful tool for the treatment of single data for detecting relative similarities rather than great differences of the absolute values of the contamination data. Mean-transect contamination levels were used for the CFA analysis in order to confirm the similarities and the differences between the transects. The CFA of the mean values of all fourteen African transects is reported in Fig. 5.

2248

HCB

Tr-Cu

S$ -Is Mr-Kin

$i-L$

CB-To

.qP~-DDT Go-At CZ-HCH

GilS~

x-HCH

B~-Bu Ma-ls

Ma-Gu qp'-DDT

Rp~DDE

/.,m-V/

Md-Ko Fig. 5 - Simultaneousrepresentationof the six chemicalsanalyzed(cx-HCH,HCB, ,/-HCH,p,p'-DDE, op'-DDT and p,p'DDT) and of the fourteen "African" transects: Sierra Leone (Si-Le), Lake Victoria (La-Vi), SeychellesIslands (Se-ls) and Mauritius Island (Ma-ls), Mali-Guinca(Ma-Gu), Benin-BurkinaFaso (Be-Bu), Ghana Suhum (Ga-Su), Ivory Coast (Iv-Co), Ghana Accra (Ga-Ac). Mount Kenya (Mt-Ke), Nairobi, Kenya (Na-Ke), Mombasa,Kenya (Mo-Ke), Cape Town (Ca-To) and Tristan da Cunha (Tr-Cu). Concentrationdata of the consideredtransects are referredto the C50 or geometricmean values of Tab. 3 and 4. The plane of the graphicalrepresentationof the CFA analysisis determinedby the secondaxis in horizontaland the third axis in vertical.

CFA can be represented by several figures in relation to the various combination of the axes which define the vectorial space (opposition of parameters and variability). From the CFA of Fig. 5, the fourteen transects appear to be divided in four clouds, indicating four different contamination patterns. Ghana Suhum is the most separated transect, characterized by a high 7-HCH level; high ot-HCH contamination levels are typical of the most remote areas, such as Tristan da Cunha, the Seychelles and Mount Kenya; the other transects, dominated by DDT contamination, are divided by the relative abundance of the original product F,p:DDT in relation to its more stable degradation product p,p'-DDE Momhasa Kenya and Lake Victoria are characterized by higher DDE values while DDT level is characteristic of Mauritius, Benin-Burkina Faso, Sierra Leone, Mali-Guinea, Ivory Coast, Nairobi Kenya, Ghana Accra and Cape Town. The selection of the axes in Fig. 5 is determined by the higher information content of the second and third axes representation in comparison to the one composed by the first and the second axes. In fact the first axis, which explains the major variability percentage (56%), discriminates the observations only for one variable (ctHCH), while the second and the third axes for two variables each: y-HCH and ct-HCH, HCB and p,p:DDE respectively. The information represented by these last two axes is more relevant, even if less striking, than that of the first and the second axis. The representation of the first and second axes, not shown, distinguishes principally two cloud of data: most of the samples opposed to Tristan da Cunha, Seychelles and Mount Kenya. These areas are the more differentiated referring to the composition of the contamination, due to their geographical position (remote areas).

2249

The relatively higher composition of ct-HCH or y-HCH seems to be the first selective criteria for the differentiation of the African contamination patterns, while the DDTs are the more abundant and diffused contaminants. [DDE]/IDDTI_ r~ti¢ in olant foliage and DDT use The contamination levels observed in the African samples are the result of many events and processes occurring when a chemical is dispersed into the environment, such as diffusion, transformation and transport phenomena. Environmental degradation can be deduced by measuring the parent compound together with different degradation products or isomeric forms. By the analysis of the data of Tab. 2, it can be observed that high contamination levels of DDT compounds (Sierra Leone and Mauritius samples), probably deriving from recent and heavy use, produce high levels of degradation products in the proportion DDT>DDD>DDE. In the case of low contamination (Seychelles and Lake Victoria), which may derive from an "old" local contamination or from a new light use or short-range transport, there are two possibilities: one with [DDE]/[DDT] ratio <1 and the other with [DDE]/[DDT] one >1. The first case is typical of light recent use or of a short-range transport from more contaminated use areas, the second one is typical of an "old" contamination status or of a long-range transport origin. The proportion of the DDT compounds, in this last case, is DDF.>DDT>DDD. The technical product of DDT is a mixture of its main derivatives, a typical example of composition of which is as follows (WHO, 1979): 77% p,p'-DDT, 15% o,p'DDT, 0.3% p,p'-DDD, 0.1% o,p'-DDD, 4% p,p'DDE, 0.1% o,p'-DDE, 3.5% unidentified compounds. Ageing causes a change in the DDT composition pattern, and the ratio [DDE]/[DDT] may be applied to identify if the found DDT is, or not, an "old" DDT. Fig. 6 shows [DDE]/[DDT] log/log plots of single samples from the four studied areas.

101110110 (d.m.) ItOOO"

log [DDE]--y log [DDT]--X

100-

LW Q Q ~J

,)

10 -

f

I-

I Qbackground/Old [] t r i n ~ l t i o n

0.1 0.1

I

sb 16o ,~.,~ " - D O T

1,6oo

Fig. 6 - [DDEI/[DDT] log/log plots of single samples from the four studied areas.

lo;ooo ~ Dmol/g (d.m.)

2250

In Fig. 6 it is shown, by means of a log/log plot of all single-sample values from the present investigation, that the [DDE]/[DDT] ratios can be grouped into three different levels: background or "old" contamination, transition and use areas. The majority of the findings from Lake Victoria (with the exception of sample LV-6, Tab. 2) and the sample SE-4 from the Seychelles Islands, are indicated by the empty squares in Fig. 6. These correspond to a typical African background or "old" contamination condition. The log/log relationship is characterized by a unitary slope (0.999), indicating that, after the elimination of log from both sides, [DDE] = 100.191 [DDT] or [DDE] = 1.55 [DDT], which corresponds to [DDE]/[DDT] =1.55. At the other extreme there are findings from recently treated areas, where the slope of the log/log is again unitary (0.991) but [DDE] = 0.15 [DDT], or [DDE]/[DDT] = 0.15. A couple of points are located in the middle of the background or "old" contamination and use areas functions, and they can be called as "transition" (LV-6 and SE-3 samples, Tab. 2), where the [DDE]/[DDT] is around 0.5. In application areas (black squares) it is possible to find different degrees of contamination, without changing [DDE]/[DDT] ratio, constant over more than 3 order of magnitude, referring to single sample data. In Fig. 7, a trial to extend this data to other African areas (including in these Tristan da Cunha) from a previous study (Calamari et al., 1991), is shown.

Dmol/g (d.w.) tO0

-

l o g [DOE] =Y l o g [DOT] =X I Y=O 191"0 ¢ = g 9 X ' ;

IJJ D O I

/

"

lo-

0~ I Y- ~ l l f*B lU" If

1

O.1

~ O BISI+I O61x Lr"=0"993 " 1

J

n background/Old

Qt r a n B I t l o n



0.01 0.1

'1

100

10

~.,z~

~OOT

umlB

BI"=IBBB

1,000

)

p-ou/=a

(d.zu.)

Fig. 7 - [DDE]/[DDT] log/logplots of C50 values from 11 Africantransects, and from Tristan da Cunha, Mauritius and S~'chelles Island The squares indicate the median concentration value (C50) for each area (from Tab. 3 and 4). Only two areas, Lake Victoria and Mombasa transect, appear as background or "old" contamination sites; due to the limited number of experimental points, the same background function as in Fig. 6 was imposed to these two points, showing a good fit; it is interesting to point out that the "overall" use area-function (slope very close to the unity) indicates a [DDE]/[DDT] ratio of 0.14. A few transition conditions can be seen, corresponding to the following transects: Ivory Coast, Nairobi and Mount Kenya. Referring to median concentration data (C50), use areas show different degrees of contamination, without changing [DDE]/[DDT] ratio, constant over more than 2 order of magnitude.

2251

CONCLUSIONS

General trends of chlorinated hydrocarbon distribution have been indicated by Calamari et al. (1991) mostly in relation to remote areas. Physico-chemical properties in combination with environmental features, e.g. cold temperatures, have been indicated as the most important factors of the remote area contamination. In the use areas the mean contamination levels are determined by two components, local/regional and global, and it is very difficult to try to distinguish the relative contribution of the two components. By analyzing the results proposed here, the most important factor determining the contamination pattern in the use areas is the history of the use of the pesticide products. The samples of the same transect reveal the same contamination pattern of the pollutants considered, showing that homogeneous zones, in terms of technological levels of agriculture and socio-economical conditions, have the same distribution patterns for the organochlorine pesticides analyzed. The typical level and composition of the several contaminants in an homogeneous zone is called the "fingerprint" of

the area. From the "fingerprint", which represents the

distribution pattern of an area, much information can be derived: a) the average intensity of use of the different pesticides; b) the age of the contamination; c) the main environmental processes which probably occurred. Environmental transformation processes can be derived from the relative contamination pattern of directcontaminated regions as opposed to surrounding indirect-contaminated areas as well as from the metabolite ratio in the same area. Heavy present and past DDT use in the majority of studied African areas is evident, as shown by the low [DDE]/[DDT] ratio (0.14) found in the majority of all considered areas, with the exception of Lake Victoria and Mombasa transects, where the ratios indicate an "old" use, probably, the former, due to the past use of DDT in the area against sleeping sickness. Intense technologically-advanced agriculture regions can be revealed by the high T-HCH contamination, as in Ghana Suhum. The Mount Kenya spectrum shows a "cold condenser effect" (Calamari et al., 1991) and the effect of proximity to more highly contaminated surrounding areas. In recent use areas, the proportion of DDT compounds is DDT>DDD>DDE, while in more indirect places it is DDE>DDT>DDD. A relatively fast anaerobic-soil degradation is possible in tropical regions revealing high DDD levels; atmospheric transformation or slow-rate soil degradation is also important as shown by the presence of DDE The Correspondence Factor Analysis or the Log-Probit elaboration will continue to play an important role in the description of the typical distribution pattern of an area or in the definition of the mean contamination level.

ACKNOWLEDGEMENTS We thank all the people that helped in the mango leaves collection. The research was supported by funds from the Italian "Ministero dell'Universita e della Ricerca Scientifica e Tecnologica, (1991 / 40%)".

2252

REFERENCES Bacci E., Calamari D., Gaggi C., Biney C., Focardi S. and Morosini M. (1988), Chemosphere, 17, 693-702. Bacci E and Gaggi C (1986), Bull. Environ. Comam. ToxicoL, 37, 850-857. Bacci E, Calamari D., Gaggi C. and Vighi M (1990), Environ. Sci. Technol., 24, 885-889. Bocci E, Cerejeira M J., Gaggi C., ChemeUoG., Calamari D. and Vighi M 0990), Chemosphere, 21, 525-535. Calamari D., Bocci E., Focardi S., Gaggi C., Morosini M. and Vighi M (1991), Environ. Sci. Technol., 25, 1489-1495. Calamari D., Tremolada P., Di Guardo A. and Vighi M. (1993), submitted toEnviron. Sci. Technol.. Carlberg G. E, Ofstad E. B., Drangsholt H. and Steinnes E. (1983), Chemosphere, 12, 341-356. Devillers J. and Karcher W. (1990), in: Practical Applications of ~uantitative Structure-Activity Relationships (QSAR) in environmental Chemistry and Toxicology, W. Karcher and J. Deviilers Editors, Kluwer Academic Publishers Group, Dordrecht, The Netherlands, 18l- 195. Gaggi C., Bocci E, Calamari D. and Fanelli R. 0985), Chemosphere, 14, 1673-1686. Nash R. G. and Beall M. L., Jr. (1970), Science, 168, 1109-1111. Pacyna J. M and Oehme M. (1988), Atmos. Environ., 22, 243-257. Paterson S., Mackay D., Bacci E and Calamari D. 0990, Environ. Sci. Technol., 25, 866-871. Risebrough R. W., Walker II W., Schmidt T T , de Lappe B.W. and Connors C.W. (1976), Nature, 264, 738739. Risebrough R. W. (1990) in: Long Range Transport of Pesticides, D. A. Kurtz Editor, Lewis Publishers, Inc. Chelsea, Michigan, 417-426. Trapp S., Matthies M., Scheunert I. and Topp E. M. (1990), Environ. Sci. TechnoL, 24, 1246-1252. Yrevors J. T. (1986), Bull. Environ. Contain. Toxicol., 37, 18-26. WHO (1979), Environmental Health Criteria 9. DD T attd its Derivatives, World Health Organization, Geneva.