Journal of Chemical Ecology. Vol. 22, No. 2, /996

SEARCH FOR TSETSE ATTRACTANTS: A STRUCTUREACTIVITY STUDY ON 1-OCTEN-3-OL IN Glossina fuscipes fuscipes (DIPTERA: GLOSSINIDAE)

W.M.

VAN C.J.

DER DEN

GOES OTTER,

VAN and

NATERS,* R.G.

L.

BOOTSMA,

BELEMTOUGRI

Department t~[Animal Physioh~gy Universi O' of Grmfingen PO Box 14, 9750 AA Haren (Gn). The Netherlands (Received May 8, 1995; accepted October 15. 1995)

Abstract--Trapping tsetse flies belonging to the palpalis group still relies totally upon luring by visual cues even though odor-baited trapping is used effectively against the morsimns-group species. Forty-three percent of the antennaI ol|:actory cells of Glossma f fitscipes, a member of the ptdpalis group, respond to I-octen-3-ol. For this species we report a structure-activity relationship between l-octen-3-ol analogs, in which carbon chain length and the configuration of the hydroxyl and r-bond moieties are varied, and biological activity. Although the optimurn chain length for all cells sensitive to l-octen-3-ol is eight and most cells give lower responses when the hydroxyl function is omitted, there is a clear division into two groups. One group is diverse and represents cells that appear indifferent to the presence or position of the r bond: many will respond to such disparate structures as acetone and 3-methylphenol as ,,yell as to l-octen-3-ol. In the other group, the structural requirements tot the stimulus are more stringent; the cells appear to be specifically tuned to l-octen-3-ol. Their thresholds are three orders of magnitude lower than those of the former group. The existence of two clusters points to a functional division in the olfactory sense. We suggest that the latter lowthreshold group is involved in host detection from a distance while the lormer diverse group is involved in host discrimination at close range. Trap harvests with I-octen-3-ol as a bait may have been disappointing because the appropriate mixture for generating a landing response on the traps is still lacking.

Key Words--Tsetse, Glossina f fuscipes, host location, olfaction, structureactivity relationship, electrophysiotogy, l-octen-3-oL *To whom correspondence should be addressed.

343 (}098-0331196/()2t)0-03435~.50/0 c, 19'46 Plenum Publishing

Ctlrp(ll'idlion

344

VAN DER GOESVAN NATERS ET AL. INTRODUCTION

The thirty-one species of tsetse flies (Glossina spp.) have been classified by morphology and habitat into three groups: the morsitans, palpalis, and fusca tsetse (cf. Jordan, 1977). All feed exclusively on blood, which they take from mammals, birds, or reptiles (Weitz, 1963). Although host odors lure flies of the morsitans group (e.g., Hargrove and Vale, 1978; Owaga, 1984; Ton-, 1989; Vale, 1974), the palpalis group "gives the impression that it detects its hosts by sight rather than smell" (ex., Laveissibre et al., 1990). Indeed, trapping of the palpafis-group flies now still relies entirely on providing visual cues (see Cuisance et al,, 1991; Gouteux and Sinda, 1990: Gouteux and Le Gall, 1992: and Ogwal et al., 1992 for recent examples). Chemical baits that attract the morsitans-group tsetse include l-octen-3-ol (Hall et al., 1984), a blend of phenols (Hassanali et al., 1986: Owaga et al., 1988; Vale et al., 1988). and acetone (Vale and Hall, t985). These are sensed by the antennae. Screening spike responses of olfactory cells to the array of attractants shows that in the morsitans-group species G. m. morsitans, 40% of the olfactory cells respond to l-octen-3-ol, 18% to the phenols, and 10% to acetone. A similar survey of the receptors in the palpalis-group species G. f fuscipes gives a surprising result: the percentages are 43 for l-octen-3-ol, 18 for the phenols, and 9 for acetone. Moreover, the sensitivity of the receptors for these odors appears to be the same in the two species (Den Otter and Van der Goes van Naters, 1992; Van der Goes van Naters, unpublished). Thus, electrophysiological experiments suggest that the organization of the olfactory sense in the two is similar even though field trials with odors indicate that the morsitans- and palpalis-group tsetse behave quite differently. Clearly, our understanding of the sensory basis of host location in the palpalis-group tsetse is still limited. Reexamination of the odor spectra of the antennal olfactory receptors, of which cells responsive to 1-octen-3-ol constitute the largest set, is needed. Our objective in this structure-activity relationship (SAR) study on octenol-derived chemicals is to characterize the specificity of olfactory receptors in G. f fuscipes and to theorize on their role in host location.

METHODS AND MATERIALS

Insects. Pupae of G, fuscipesfuscipes Newstead 1910 were obtained from the colony of the Drpartement d'Elevage et de Mrdecine Vrtfrinaire (CIRADEMVT) in Montpellier, France. The pupae were kept individually in glass vials until emergence of the flies, whereupon these were grouped in cages by sex and age. Pupae and flies were kept in a 12L: 12D photoperiod at 25°C and 75%

SAR STUDY ON TSETSE ATTRACTANTS

345

relative humidity. Flies were fed off rabbits' ears on alternate days beginning the day after emergence. Etectrophysiology. Recording of action potentials was accomplished as described by Den Otter and Van der Goes van Naters (1992). A fly was tested when between 4 and 15 days old; it was fed the previous day. Cells at two sites on the antennae of each fly were tested, one on the medial funicular face and the other on the lateral face. A current of air, issuing from a 7-mm-ID tube at 2 m/sec, was directed at the head of the fly. The tube was provided with an opening through which test compounds were injected. Test Compounds. Test chemicals used included 1-octen-3-ol and structurally related compounds in three sets of experiments (Table 1). These chemicals vary in carbon chain length and in the positions/presence of the ~r bond and hydroxyl moieties. Additionally, acetone and 3-methylphenol were tested. Purity according to specification was > 98 % and chiral substances were racemics unless otherwise indicated. All were purchased from Fluka (Buchs, Switzerland) except for the three isomers 2-, 3-, and 5-octen-l-ol (Johnson Matthey, Karlsruhe, Germany), 1-hexen-3-ol (Aldrich, Milwaukee, Wisconsin), and l-nonen-3-ol (ICN, Cleveland, Ohio), These chemicals are liquids at room temperature and pressure (RTP: 298K, 1 atm). Saturated vapor concentrations at RTP, as determined by weight loss over 24 hr of samples in glass-stoppered 1-1iter flasks, are given in Table I. Three log step dilutions in air were prepared by transferring appropriate amounts of the saturated vapors of the chemicals in 1.2-1iter glass serum bottles fitted with screw caps and rubber seals. Saturation of the atmosphere in a "mother" bottle was achieved and maintained throughout the study by addition of 2 ml of a chemical. The three dilutions from these were remade at the beginning of each day: two dilution series of bottles for each substance were alternated. Thus a bottle used on one day was left standing uncapped the next, the cap and seal being placed on its twin. This allowed the uncapped bottle to cleanse of odor while the rubber seal, after a number of exchanges, had absorbed what it could and was no longer a biasing factor on the concentration of a dilution. Test chemicals were drawn from the bottles through the rubber seals into plastic syringes (Omnifix 5 ml) immediately before presentation. Syringes were equilibrated with the bottle atmosphere by pumping the plunger 10 times prior to taking a sample. Not more than l0 syringe fills were taken from a bottle per day. A syringe was mounted on a spring-powered device and 1.5 ml of its contents was injected in 0.1 sec into the airstream directed at the head of the fly, Vapors of acetone and 3-methylphenol were not taken from bottles but were dispensed from filter papers placed inside syringes, Syringes with acetone

VAN DER GOES VAN NATERS ET AL.

346

TABLE 1. CHEMICALS USED IN EXPERIMENTS 1-3"

Saturated

Name I-octen-3-ol

Structure

vapor conc J'

~

7.6 × 10 ~'M

OH 1) l-propen-3-ol

~

1.6 x 10 ~M

/

OH

I-buten-3-ol

~"N/ " i

l-penten-3-ol

~

7.6 x 10 4M

OH 4.5 x 10 ~M

/

OH l-hexen-3-ol

~

l-heplen-3-ol

~

W

t,4 x 10 4 M

OH ~

6.4 x 10 S M

/

OH l-nonen-3-ol

~----~x~,

~

7.1 x 10 " M

OH 2) 2(E)-octen-I-ol

~

~

/

3.2 x 10

~

M

( 31Z)-octen-l-t~l

0 7 ~ / / ~ ,

3.3 x 10 "M

OH 5tZ)-octen-l-o~

i~

N

~

~

3.5 x lO " M

/

7.9 x 10 ~ M

OH 3) octane

/

/

l-octene

~'x

3-octanol

./

~

~

/',,

/',

/

~ N / V ",T/ V

V

7.6 x 10 4 M

7.5 x 10

OH "The parent compound l-oclen-3-ol was the relerence chemical in each experinaent.

J'AI RTP: 298K, I atm.

contained 10 mg o f the chemical; those with 3-methylphenol were charged with 5 mg. Experimental Procedure, Cells on the antennae were screened for response to a stimulus o f saturated 1-octen-3-ol vapor. W h e n a cell sensitive to 1-octen-

347

SAR STUDY ON TSETSE ATTRACTANTS

3-OI was located, the test chemicals were drawn from the bottles and injected in ascending concentration. The order in which substances were presented was varied. We observed intervals of 30 sec between presentations to allow the cells to return to baseline activity. After all concentrations of each substance in an experiment had been presented, the odor spectra were further characterized with acetone and 3-methylphenol. Analysis of Data. Recordings were transferred from tape to paper for analysis. The magnitude of response was calculated from the maximum number of action potentials in a space of 0.1 sec after injection of the stimulus, the nonstimulated activity prior to presentation of the odor being similarly measured and subtracted. This figure was multiplied by 10 to obtain spikes per second as the unit of response. To characterize the odor spectra of the olfactory receptors, multidimensional graphs were envisaged with each axis representing the response to a different substance relative to that on stimulation with l-octen-3-ot. Here we present graphs with two dimensions only, as more are difficult to visualize. Cells with similar spectra form clusters. As a supplement to this common-sense visual approach for grouping cells, several types of multivariate statistical analyses were considered, including hierarchical cluster analysis, factor analysis, and multidimensional scaling. Bieber and Smith (1986) note that for cluster and factor analysis "'the popularity of these intuitive procedures is rooted in the absence of sound statistical theory concerning the appropriateness of the final solution." Multidimensional scaling has the characteristic that the meaning of the dimensions that reveal the underlying structure in the data is abstract. We have opted for the first of the three (hierarchical cluster analysis from SPSS/PC + with average linkage over Euclidean distances) and base our choice of cluster solution on a scree procedure. The different clusters that emerged from the analyses were considered separately, and the average responses of the constituent cells are plotted in doseresponse curves. Points in the curves are connected by line segments; sigmoid or other curve fitting did not appear feasible in our graphs due to the limited number of concentrations per substance. This has a drawback: statistical analysis of the dose-response curves is not possible without a function to fit to the data because the concentrations used are different for each chemical. Therefore conclusions are based on the most salient features of the data only. RESULTS The results comprise a data set of 183 cells recorded from 41 female flies and 52 males. Responses to four concentrations of 1-octen-3-ol did not differ between the sexes (ANOVA, P = 0.84); hence the data are pooled. The three experiments (Table 1) are treated separately.

348

V A N DER G O E S VAN N A T E R S ET AL.

Effect of Chain Length. T h e relative effectiveness o f the six h o m o l o g s o f 1-octen-3-ol is the same for all 28 cells tested, as is reflected by the graph in Figure la. Figure l a s h o w s a single cluster in a plot with the relative responses to l-hepten-3-ol and 1-nonen-3-ol on the axes. This is true also in graphs with a) 3

c)

A2 E

n=28

n=28

08

200

og

~c7

150

P-) r

E-

g

I00

c

-3

-2 -1 0 1 2 t-hepten- 3-ol/

3

l-octen- 3-oi

10-

10.8 10-'~ 10-6 lO-s ]0 4 10-] concentration (N)

b) c eo

i mI

¢i

i Ill•HI

0

].2 ]0

8

6

4

2

Number of Cqusters

FiG. 1. (a) Spike responses of cells on stimulation with t-hepten-3-ol (abscissa) and l-nonen-3-ol (ordinate) relative to the responses on stimulation with the parent compound 1-octen-3-ol, Each data point represents one cell. The plotted values are proportions, have no units, and are averages over four concentrations. Axes are numbered with exponents and extend from 10 .5 to 103. Stimuli to be graphed on the axes were selected to maximize clustering of cells. The data points are seen as a single cluster. (b) Scree diagram for the last 12 agglomeration stages in the hierarchical cluster analysis on the 28 cells. Dimensions in the analysis are the relative responses to the six homologs of locten-3-ol as in (a). The ordinate gives the distance over which the two nearest clusters are combined to create the number of clusters given on the abscissa. Each successive stage fuses two clusters that are further apart. The rise is steady and the distances are small as compared with Figure 2b. We choose the one-cluster solution. (c) Dose-response curves for l-octen-3-ol (C8) and its homologs. The abscissa is the molar concentration of the stimulus; the ordinate gives the average spike responses with SEM of the cells in the cluster of (a).

SAR STUDY ON TSETSE ATTRACTANTS

349

any other combination of the homologs represented (not shown). When cells are grouped by hierarchical cluster analysis, the Euclidean distances over successive stages of the agglomeration process show a steady increase (Figure I b): we allow clustering to progress to its end. In the sample, five cells responded to acetone, three to 3-methylphenol, four to both, and 16 to neither. Figure lc shows the average responses to the saturated vapors and three log step dilutions of each homolog. Note that the curves are staggered on the abscissa: the saturated vapor pressures decrease with increasing chain length (Table l). The C7 and C9 curves overlap. A chain length of eight carbon atoms is clearly optimal in the range of homologs tested. Effect of Relative Positions ~r-Bond and OH Moieties. When the relative responses to 2-octen-l-ol and 3-octen-l-ol are represented on the axes, two diffuse clusters are apparent (Figure 2a: arbitrarily designated P and Q). These represent 19 and 32 cells, respectively. The scree procedure confirms that a two-cluster solution is appropriate (Figure 2b). Recordings from the cells in cluster P always showed two spike amplitudes, which suggests that they are collocated with a second type of cell (Figure 4). Further investigation proved the latter responds to 3-methylphenol only and not to any of the other substances tested. Cluster P cells themselves are unresponsive to either acetone or 3-methylphenol. The dose-response curves show that the cells are very discriminating to deviation of the positions of the ~r-bond and hydroxyl moieties (Figure 2c). The 1-3 configuration is the only adequate structure. Of the cells in cluster Q, 8 responded to acetone, 2 to 3-methylphenol, 10 to both, and 12 to neither. The curve of 1-octen-3-ol (Figure 2d) lies somewhat below the other three. This suggests that positioning of the hydroxyl group on carbon 1, instead of on carbon 3, may enhance the response. A closer look at Figure 2a seems to bear this out. The points in the cluster are spread over a 45 ° diagonal relative to the two axes: the variation in the cluster is largely due to the OH-group position rather than to the position of the a- bond. Although cells in the left range of cluster Q appear to be indifferent to the OH position (relative responses are around 10° = I), there is a continuum up to cells that respond 10 times more strongly on stimulation with 2- or 3-octen-l-ol than with 1-octen3-ol. The cells seem indifferent to the position of the 7r bond (Figure 2d). Due to a procedural error, the lowest point in each curve was lost. Effect of Omission of 7r Bond or OH. Cluster formation is less pronounced in this experiment (Figure 3a). Several cells are detached from two rather dense agglomerations. This is reflected in the scree diagram by a less salient elbow point (Figure 3b) than is shown in Figure 2b. For ease of interpretation, we chose a two-cluster solution. These are again assigned letters P and Q (38 and 66 cells, respectively).

350

VAN DER GOES VAN NATERS ET AL,

a)

c) 31

P n;19

300

+ 21

E

c

eo

1-octen- 2-oi

-'~ 200

/

150

o:32 o

-2i 4 . Pnq9

~

-3 ..................

2 - o c t e n 1-ol

so

.>oete~->ol

0 ......

-3 -2 -I 0 I 2 3 3-octen-I-of/ l-octen-3-ol b) •

C

E tr~

~

200~

-~ 2, c I

~ ~

150

[

> i iii,,

,i"

• • •

.................... 0~2

I0

8

6

4

number of clusters

i

10 - 5

10 - 4

i0 -3

.L 3 - o c t e n - l - o t I 2-octen-l-ol

~

8 loo! ~ I

~'~

o: - - - 2

t e~Q£-nl'-°! ~ 10 - 6

300~ O n::32 250

2vq

-

c 10 . 7

concentrabon (M) d)

8 3~

~

i0 -9 20.8

10 - 9

5-octen-l-ol 5 1-octen-3-o[

"

_t_a£]_ . . . . . . . . . . . . . . . . . . . . . .

10 - 8 l0 - 7 l0 - 6 10 - 5 l0 - 4 i0 - 3

concentrabon (M)

FIG. 2. (a) Spike responses of cells on stimulation with 3-octen-l-ol (abscissat and 2-octen-l-ol (ordinate) relative to the responses on stimulation with t-octen-3-ol. See Figure la for details. The two clusters are marked P and Q. (b) Scree diagram for the hierarchical cluster analysis by relative response to 2-, 3-, and 5-octen-l-ol. See Figure lb for details. A salient elbow in the steady rise of Euclidean distances at the two cluster solution suggests agglomeration should be terminated there. (c) Dose-response curves lbr the cluster P cells in (a). Abscissa: molar concentration of the stimulus; ordinate: average spike responses with SEM. (d) Dose-response curves for the cluster Q cells in (a). Abscissa: molar concentration of the stimulus; ordinate: average spike responses with SEM.

Recordings of cells in cluster P display the characteristics described in Figure 4. The curves in Figure 3c show that l-octen-3-ol is clearly the most stimulating followed by 3-octanol and 1-octene. Octane gives little or no response. Cluster Q has a wider spread. In this cluster, 16 cells responded to acetone, 8 to 3-methylphenol, 17 to both, and 25 to neither. The curves for octane and 1-octene overlap, as do the pair for 3-octanol and 1-octen-3-ol (Figure 3d). The latter are clearly more stimulatory on average. However, the averages in Figure 3d mask the extremes. Thus, for a number of cells of the cluster (those on the

SAR STUDY ON TSETSE ATTRACTANTS

a)

35 l

c)

3

l-octen- 3-ol

n : 38 _

200f

~

3 - o c t anol

ol •. ~o 1ooi o

:

:

Pn:38

_3 t . . . . . . . . . . -3 -2 -1 0 1 2

I

octane

10 - 9

3

10 - 8

b)

lO - 7

10 - 6

10 - 5

10 4

10-~

concentrat,on (M)

octane / i-octen- 3 - ol

d)

3[

n:66

~c

ooi 1501.

•

g

1oo}

ff

~o f

1-octen-3-ol 3-octanol

.~

i : octane :-octene

~

I mlmnmlml

I I2 10 8 6 4 2 number of d u s t e r s

,o°!o 1o:8

~

/

1o-4 concentration (M)

FIG. 3. (a) Spike responses of cells on stimulation with octane (abscissa) and 3-octanol (ordinate) relative to the responses on stimulation with l-octen-3-ol. See Figure la for details. For ease of interpretation, five outliers are disregarded to give two clusters, P and Q. (b) Scree diagram for the hierarchical cluster analysis by relative response to octane, t-octene, and 3-octanot. See Figure la for details. We choose the two-cluster solution. (c) Dose-response curves for cluster P ceils in (a). Abscissa: molar concentration of the stimulus; ordinate: average spike responses with SEM. (d) Dose-response curves for cluster Q cells in (a). Abscissa: molar concentration of the stimulus; ordinate: average spike responses with SEM.

I 1mY 0.1s FIG. 4. Response of a cluster P cell (large-amplitude spikes) to a low stimulus intensity of 1-octen-3-ol (0. l-see duration). Cluster P cells continue firing at an elevated rate beyond stimulus duration. They are insensitive to acetone and 3-methylphenol. Collocated always is a phenol-sensitive cell with similar temporal characteristics (small-amplitude spikes).

352

VAN DERGOESVAN NATERSET AL.

right-hand side in Figure 3a), the responses to octane and 1-octene are 10 times those to 1-octen-3-ol.

DISCUSSION

The sample of 183 cells is a motley conglomeration. An underlying order can, however, be made apparent even though our tools are crude. The grouping of cells here is not an expression of our human inclination to classify but is a real feature of the data set. The existence of clusters in the population of cells responsive to 1-octen-3-ol points to a functional division in the olfactory sense. Clustering proves that the population of olfactory cells works with more than one type of receptor molecule. The wide spread of data points within the clusters--especially for clusters Q in Figures 2a and 3a--suggests that even within a subset of the population there are different types. This may extend to the cell membrane, as was shown for the sugar cells of the fleshfly (Shimada et al., 1974). Until we are sure that a membrane harbors but one type of receptor site, we cannot model sites from SAR studies on a cellular level. The most we can do is indicate which moieties in an odor molecule play a role in the odormembrane interactions. Clustering also shows the importance of testing the olfactory cells individually. Antennal differentiation necessitates that reception of odors be studied at the sensory level, by single-cell recording if possible. Perception of odors is reflected by the flies' responses in wind tunnels and field trials. Other behavioral assays may give valuable data as well, although they can be difficult to interpret. Saini et al. (1989) reported an SAR study on l-octen-3-ol in which they assessed the relative perception of odors by counting the number of flies that raise their antennae to a stimulus. They concluded that the chemoreceptors involved may not be very specific and commented on features of the receptor sites. The differences between our results may be a contrast in species--their work is on G. m. m o r s i t a n s - - b u t more likely reflect differing methodologies. Whether or not cells form disjunct groups in Figures la, 2a, and 3a seems to be dependent on the type of chemicals tested. While all cells respond in similar fashion to the homologous series 1-propen-3-ol through 1-nonen-3-ol, they are differentially sensitive to the other 1-octen-3-ol derivatives. Clusters P in experiments 2 and 3 (Figures 2a and 3a) appear to correspond: cells of both have the characteristics outlined in Figure 4. From this it follows that clusters Q in experiments 2 and 3 (Figures 2a and 3a) must also be identical. That the two clusters P each represent the same proportion of the total, approximately 37% (experiment 2:19/51 ~- 37.3%; experiment 3:38/104 ~ 36.5%), supports this. The thresholds of cluster Q cells for 1-octen-3-ol are on average between

SAR S T U D Y ON T S E T S E A T T R A C T A N T S

353

10 -s and 10 -7 M (Figures 2d and 3d). The position or presence of a rc bond in the molecules appears to be of no consequence. The hydroxyl group is important (Figure 3d: the curve for 3-octanol intercepts the abscissa > 1 log unit lower than the octane curve), and the effect of its position on the spike responses varies between nil and a 10-fold factor for the substances tested (Figure 2a). On average, molecules appear more stimulatory with an OH on carbon 1 than on carbon 3 (Figure 2d). For cluster P cells the structural requirements for a good stimulant appear to be more stringent than for the cluster Q cells, Both the presence of the 7rbond and hydroxyl moieties as well as their position in the molecule are important (experiments 2 and 3: Figures 2c and 3c). None of the tested compounds meet the requirements as well as I-octen-3-ol. Extrapolating the dose-response curves for 1-octen-3-ol in Figures 2c and 3c to the abscissa gives an average threshold concentration of 10 -~2 to 10 -~° M. According to the latest data now available (Torr et al+, personal communication), an ox exhales 0-0.119 mg 1-octen-3-ol/hr with an average of 0.014 rag/hr. Presupposing a ventilation rate of approximately 5000 liters/hr (Spector, 1956) and with a relative molecular mass of 128.22 g/tool for l-octen-3-ol, its concentration in cattle breath is 2 × 10 -it M [0.014 × 10 3/(128.22 × 5000) g/mol/hr/(hr/g/liter)]. Downwind, the odor spreads into a larger volume. although the rate of diffusion is difficult to estimate. Despite the decline in concentration downwind of an ox, we suggest that cluster P cells are sufficiently sensitive to l-octen-3-ol to aid in host detection. Furthermore, threshold concentrations between 10 I-~and 10-~o M will be overestimates if the fly applies a more sensitive method of decoding its afferent signals than we have used in our analyses of the recordings. It is clear, however, that the cells will be functioning at the lower limit of their dynamic range. The cells appear to be specifically tuned to 1-octen-3-ol; no structural alteration tested augmented the response nor did structurally unrelated chemicals prove effective. The function of the cluster Q cells remains open due to their high thresholds. Many of these cells respond to structurally unrelated compounds as well as to l-octen-3-ol and its derivatives. Either the prime stimulant is not among these or chemicals in mixtures may act synergistically on a cell. The latter possibility is now being investigated. With dissimilar but overlapping odor spectra, the cluster Q cells could be involved in discriminating the odor mixtures from different hosts (across fiber pattern coding; cf. Erickson, 1963, 1982). We suggest the cluster P cells aid in host detection from a distance while the cluster Q cells are involved in host recognition at close range. Contrary to the results of the trapping experiments in the field, our electrophysiological experiments indicate l-octen-3-ol may be a powerful attractant for G. f fuscipes. Sutcliffe (1987) notes that disparate odors mediate successive

354

VAN DER GOES VAN NATERS ET AL.

stages in the distance orientation of biting flies to their hosts. The odors that attract tsetse appear to be different from those that cause the flies to alight for feeding (Saint et al., 1993), We suggest that although the palpalis-group tsetse should be attracted by 1-octen-3-ol in field tests, the trap harvests thus far may have been disappointing because the mixture required for an alighting response has not yet been developed. Acknowledgments--We are very grateful to Dr. D. Cuisance of the CIRAD-EMVT: he was so kind to again provide us with an ample supply of tsetse. We also would like to acknowledge Dr. D.G. Stavenga, who proofread the manuscript, and Dr. H. van der Wel for discussions. We thank the anonymous referees for their constructive critique. The study was financially supported by the Dutch Foundation for the Life Sciences (NWO-SLW, grant 811-425-083).

REFERENCES BIEBER, S,L., and SMITH, D,V. 1986. Multivariate analysis of sensor?,, data: A comparison of methods. Chem. Senses 11:19-47. CUISANCE, D., CAILTON, P., KOTA-GUINZA, A,, NDOKOUF-, F.. POUNI~KROZOU. E., and DEMBA, D. 1991. Lutte contre GIossina fuscipes fuscipes par pirgeage chez les ,61eveurs Mbororo de Rgpublique Centrafricaine. Rev. Elev. Med. Vet. Pays Trop. 44:81-89. DEN OTTER, C.J., and VAN DeR GOES VAN NATF.RS, W,M. 1992. Single cell recordings from tsetse (Glossina m. morsitans) antennae reveal olfactory, mechano- and cold receptors. Physiol. EntomoL 17:33M.2. ERICKSON, R.P. 1963. Sensory neural patterns and gustation, pp. 205-213, in Y. Zottennan ted.}. Olfaction and Taste. Pergamon Press, Oxlord. ERICKSON, R.P, 1982. The across-fiber pattern theory: An organizing principle for molar neural function. Contrib. Sensor3" Physiol. 6:79- I t0. GOtJTEUX, J.P., and LE GALL, F. 1992, Pi~ge bipyramidal '~ tsets6 pour la protection de FOlevage en Rrpublique Centrafricaine, CIRAD-GERDAT-PRIFAS. pp. 109-116. GOt.~TEUX, J.P., and SINDA, D. 1990. Community participation in the control of tsetse flies, Large scale trials using the pyramid trap in the Congo. Trop. Med. Parasitol. 41:49-55, HALL, D.R.. BEEVOR, P.S., CORK, A., NESBITI. B.F., and VALe=, G.A. 1984. A potent olfactory stimulant and attractant for tsetse isolated from cattle odours, btsect Sci. Appl, 5:335-339. HARGROVE, J.W., and VALE, G,A. 1978. The effect of host odour concentration on catches of tsetse flies (Glossinidae) and other Diplera in the field. Bull. Entmnof. Res. 68:607-612, HASSANAU, A., McDowF.t.L.P.G., OWAGA. M . L A . , and SAtNI. R K . t986. Identification of tsetse attractants from excretory products of a wild host animal, Svncerus cq[[er. Insect SoL Appl. 7:5-9, JorDan, A.M. 1977, Systemalics, pp. 13-22. in M. Laird {ed,). Tsetse: the Future for Biological Methods in Integrated Control. International Development Research Centre. Ottawa. LAVEISSIERE, C., EOUZAN, J.-P.. Grt-BAUT, P., and LFMASSOn, J.-J. 1990. The control of riverine tsetse, h~sect Sci. Appl. I 1:427-441. O~WAL, L,M., KANCWAC;YE,T,N., SEMAKULA, L., NDYABAmKA, C., and Estru, P. 1992. Tsetse and trypanosomiasis control lot the rural development of Buvuma Island, Lake Victoria, Uganda. Proceedings of the IAEA-FAO symposium in Muguga, Kenya. IAEA-TECDOC-634. p. 247. OWAGA, M,L.A. 1984, Preliminary observations on the efficacy of olfactory attractants derived from wild hosts of tsetse. Insect Sci. Appl. 5:87-90.

SAR STUDY ON TSETSE ATTRACTANTS

355

OWAGA, M.L.A., HASSANALI, A., and McDOWELL, P.G, 1988. The role of 4-cresol and 3-npropylphenol in the attraction of tsetse flies to buffalo urine, hTsect Sci. Appl. 9:95-100. SAINI, R.K., HASSANALI,A., and DRANSFIEI_D, R,D. 1989, Antennal responses of tsetse to analogues of the attractant I-octen-3-ot. Physiol. Entomol. 14:85-90, SAiNt, R.K., HASSANALI,A., AHUYA, P., ANDOKE. J., and NYANDAT, E. 1993, Close range responses of tsetse flies GIossina morsitans morsitans Westwood (Diptera: Glossinidae) to host body kairomones. Discoveq~" hmovatirm 5: 149-153. SHIMADA, [., SI-IIRAISHI,A., KIJIMA, H., and MORJ'rA. H. 1974. Separation of two receptor sites in a single labellar sugar receptor of the fleshfly by treatment with p-chlormercuribenzoate. J. h~sect Physiot. 20:605-621. SPECTOR, W.S, (ed.) 1956. Handbook of Biological Data. W.B, Saunders, Philadelphia. SUTCLIFFE, J.F. 1987, Distance orientation of biting flies to their hosts. Inse¢'t Sci. Appl, 8:611616, TORR. S.J. 1989. The host-orientated behaviour of tsetse flies (Glossina): The interaction of visual and olfactory stimuli. PJ~ysiol. Entomol. 14:325-340. VALE, G.A. 1974. The responses of tsetse flies (Diptera, Glossinidae) to mobile and stationary baits. Bull Entomol. Res. 64:545-588. VALE, G,A., and HAt.L, D.R. 1985. The role of l-octen-3-ol, acetone and carbon dioxide in the attraction of tsetse flies, Glossina spp, (Diptera: Glossinidae), to ox odour. Bull. Entomol. Re.z. 75:209-217. VALE, G.A., HALL, D R . , and Gotx;H, A , J E . 1988. The olfactory responses of tsetse flies, Glossina spp. (Diptera: GlossinidaeL to phenols and urine in the field. Bull. EntomoL Rex. 78:293300. WEITZ, B. 1963. The feeding habits of Gh~ssina Bull. W.H.O. 28:711-729.