Second generation of grape berry moths, Lobesia botrana (Den. & Schiff.) (Lep., Tortricidae) and Eupoecilia ambiguella (Hb.) (Lep., Cochylidae): spatial and frequency distributions of larvae, weight loss and economic injury level

J . Appl. Ent. 122. 361-368 (1998) 1998, Blackwell Wissenschafts-Verlag, Berlin ISSN 0931-2048

Second generation of grape berry moths, Lobesia bofrana (Den. & Schiff.) (Lep., Tortricidae) and EupoeciDa ambigueh (Hb.) (Lep., Cochylidae): spatial and frequency distributions of larvae, weight loss and economic injury level F. Pavan', V. Girolami' and G. Sacilotto* 'Dipartimento di Biologia applicata alla Difesa delle Piante, Universita di Udine, Italy; 'Istituto di Entomologia agraria, Universita di Padova, Italy

Abstract: Spatial and frequency distributions of grape berry moth larvae were studied in vineyards of northeastern Italy. The larval density varied in relation to position of grape clusters within vines. For density values below 0.5 larval nests per cluster, the larvae were almost randomly distributed and fit the Poisson distribution, while above 0.5 larval nests per cluster were slightly aggregated and fit the negative binomial distribution better. A new method to estimate cluster weight loss cause by second generation of grape berry moths, based on weighing and counting of berries injured, was proposed. Therefore an estimate of economic injury level based on this method was put forward.

1 Introduction

berry moths for the larger and tighter clusters (SACILOTTO,1981). Although estimates of weight loss are Second generation larvae of grape berry moths, Lobesia difficult to demonstrate by weighing the total yield or botrana (Den. & Schiff.) and Eupoecilia ambiguella cluster, missing berries associated with grape berry (Hb.) bore into clusters of berries, causing weight loss moth damage clearly show that there is some loss in and grey mould (Botrytis cinerea Pers.) damage cluster weight. (ROHERICHand SCHMID,1979; PAVANet al., 1987, The difficulties of estimating losses by weight yield 1989a, 1993a; PAVANand SBRISSA,1994; SCHRUFTet led to a proposal in Italy of a new method based on the al., 1988). Estimates of weight losses caused by the weight loss of damaged berries. VALLI(1975) estimated second generation of grape berry moths have been based the direct losses arising from an average of the three on weighing of yield (or clusters) or weighing and coun- berries attacked per larva of E. ambiguella and the subting of berries. sequent losses due to botrytis-infested berries being in Insecticide treatments against second generation of contact with tunnelled ones. On the basis of weight loss, grgpe berry moths significantly increased yield only economic injury levels (the population level that caused when infestations were high (above 50% of clusters an economic loss equal to the cost of control) of 20% infested) (ROEHRICH,1978; SCHRUFTet al., 1988; of infested clusters in Italy (VALLI,1975) and between MARTINSet al., 1989). An average yield loss of 0.07- 4% and 7% in Germany (KASTand MUNDER,1990) 0.13% per larva per 100 clusters was observed for E. were proposed for the second generation of grape berry ambiguella in Germany, but the correlation between the moths. number of larval nests per cluster and the percentage of The percentage of infested clusters is easier and yield loss of several vineyards was not always significant quicker to estimate than the number of larval nests (KAST and MUNDER,1990). The variability of the per cluster. Therefore, it is preferable to correlate the potential yield (number and weight of clusters) and the percentage of yield loss to the percentage of infested incidence of botrytis damage in observed vineyards do clusters. For this purpose it is necessary to study the not account for the weight loss caused by grape berry frequency distribution of larvae on clusters to undermoths. Even taking the average weight of clusters stand the relationship that links the percentage of instead of yield does not improve the reliability of using infested clusters to the number of larval nests per cluster. this relationship (SACILOTTO,1981; MARTINSet al., The aims of our work were: (1) to determine the 1989). spatial distribution of grape berry moth larvae in vineComparisons between the average weight of infested yards and within vines, (2) to identify the frequency and uninfested clusters showed, in one case, a loss of distribution of grape berry moth larvae on clusters, (3) 0.1% per larva per 100 clusters (KASTand MUNDER, to improve VALLI'Smethod for estimating weight loss (1981), (4) to 1990) and, in another case, an increase in comparative as previously reported in GIROLAMI weight of infested clusters due to the preference of grape empirically evaluate the new method proposed for estiu. S. Copyright Clearance Center Code Statement:

093 1-2048/98/2207-0361 $ 14.00/0

F. Pavan. V. Girolami and G. Sacilotto

362 mating weight loss, a n d (5) t o devise an estimate of the economic injury level based on t h e new method for estimate of weight loss.

2 Materials and methods

2.2 Frequency distribution Two different approaches were used to study the frequency distribution of second generation larvae on clusters. Firstly, the relationship between the mean and variance using the 1961) was investigated: Taylor power law (TAYLOR, =a

+ b log(x)

2.1 Spatial distribution

lOg(S’)

Between 1980 and 1986, in 15 different localities in northeastern Italy, 36 uniform plots (same cultivar and age, and very similar agronomic conditions) of early, middle or lateharvested cultivars (e.g. Chardonnay, Merlot, Tocai friulano, Corvina veronese) untreated with insecticide and grown with two different training systems (Sylvoz, a vertical training system, or ‘pergola veronese’, a horizontal training system) were sampled to study the spatial distribution of second generation larvae between rows and within vines (table 1). The sampling was done about 1 month after the last captures in pheromone traps. The sampled clusters were chosen on the basis of an a priori scheme to avoid a subjective choice [see PAVANet al. (1987, 1993b) for Sylvoz and PAVAN and SBRISSA (1994) for ‘pergola veronese’]. A number varying between 12 and 50 vines per vineyard was used according to fixed schemes (from 2 to 4 rows per vineyard and from 6 to 25 vines per row chosen in determined positions along the rows). For each vine one cane was considered. On each cane two shoots were chosen (the basal and the distal). On each shoot two clusters were examined. If the two shoots had less than two clusters on each, the shoot next to the basal one or the shoot previous to the distal one with a t least two clusters was examined. If more than two clusters per shoot were present, the basal clusters were examined.

where R = number of larval nests per cluster; S’ = variance. Secondly, the frequencies observed in the field were compared with the expected ones according to the classical x2 method on the basis of Poisson and negative binomial distributions; the ‘dispersion’ parameter ( k ) was calculated with the maximum likelihood method (BLISSand FISHER,1953). Between 1980 and 1994, in three different localities in northeastern Italy, 94 uniform plots (early, middle or late-harvested cultivars) untreated with insecticide and grown with two different training systems (Sylvoz or ‘pergola veronese’) were sampled to estimate the larval infestation (table I). The samplings were taken as discussed previously (see Spatial distribution). In some vineyards only L. botrana larvae were present; in others E. ambiguellu larvae were also present (table 1). The data were elaborated considering both species larval nests together, since during the samplings the larvae could be absent and it was not possible to identify the species (when specific observation were made in vineyards with both species present the percentage of E. ambiguella larvae varied from 2% to 25% of the total).

2.3 Weight loss and economic injury level Between 1979 and 1984, on two farms with different cultivars (12 cases in all), the number of larval nests per cluster, the percentage of infested clusters, the number of damaged berries

Table 1. Summarized scheme of vineyards sampled to study spatial and frequency distribution of larvae on clusters

Year

Spatial distribution data

Frequency distribution data

*

* *

Training systems and number of observed vineyards Sylvoz Pergola

Number of observed clusters

Only Lobesia botrana

observed

Both grape berry moths observed

~

1980 1981 1982

* *

*

* * *

1983 1984 1985

* *

* 1986

* * * * *

* 1987 1989 1991 1992 1993 1994

* * *

* * * * *

* * *

* Total

1 2 15 12 1 5 1 2 3 7 1

4 5 1 1 4 2 3 3 2 1 2 12 3 1

85

9

100 96 64 48 160 100 100 100 200 100 100 200 100 96 200 200 192 100 200 100 200 128 100 150 200

* * *

* *

* *

*

*

*

* * *

*

* * *

*

* * * * * *

* 26

68

Second generation of grape berry moths

363

per larva, the weight of damaged and undamaged berries, and the weight of clusters were estimated. Data were obtained observing a number of clusters varying between 48 and 160; the samplings were peformed as described previously (see Spatial distribution). The damaged berries included the ones that had been tunnelled (unrotten, rotten, turgid or shrivelled) or not (rotten in contact with tunnelled berries) as reported in PAVANet al. (1987, 1993a). In five cases the damaged berries were distinguished per each larval nest as well as per each cluster; only in these cases was the variance of the number of injured berries per larva calculated. The average weight loss of damaged berries was estimated by comparing the weights of damaged and undamaged berries. For this purpose, four groups of 30 berries, damaged and undamaged, were weighed. Undamaged berries were collected around damaged berries. The average weight of the clusters was always estimated from 100 clusters chosen with the same scheme previously described (see: Spatial distribution). For estimates of the economic injury level, yield, cost of control and price of grapes were further considered.

1

-2

I

,

-2

1

0

-1

log (mean)

3 Results and discussion

Fig. 1. Regression between log mean and log variance of larval nests per cluster

3.1 Spatial distribution

There were no significant differences in larval infestation between the rows of uniform vineyards (table 2). The shoot position within a vine influenced the level of infestation of clusters (table 2). With the Sylvoz training system the distal shoots were always significantly more frequently infested than the proximal ones. This may be because the clusters on distal shoots are bigger than the others (SACILOTTO,1981) and have a thicker and more abundant leaf covering that provides more favourable conditions for female oviposition (VALLI, 1975). With the ‘pergola veronese’ training system, the distal shoots were significantly less infested than proximal ones only in 1986. In this case the size of clusters was not significantly different (unpublished data), but the leaf covering varied due to the absence of contact between the top parts of the ‘pergolas’ of two adjacent rows. The basal or distal position of the cluster on shoots did, not significantly influence the population level (table 2).

3.2 Frequency distribution The linear relationship between mean and variance of larval nests per cluster when using log-log trans-

formation was statistically significant (r = 0.98; n = 94; P = < 0.01) with intercept a = 0.096 and slope b = 1.093 (fig. 1). For less than 0.5 larval nests per cluster, the larval population was almost randomly distributed (a = 0.030 and b = 1.014); for more than 0.5 larval nests per cluster seemed aggregated (a = 0.124 and b = 1.325). Significant differences in slope (b) for the GT2-method (HOCHBERG, 1974) were not observed between different training systems (a = 0.087 and b = 1.087 for Sylvoz and a = 0.153 and b = 1.1 19 for ‘pergola veronese’). There also were no significant differences in slope for larval populations of L. botrana alone (a = 0.1 11 and b = 1.107) or where small percentages of E. ambiguella larvae were also present (a = 0.092 and b = 1.088). Because the fitting of data to the Poisson distribution and to the negative binomial distribution can not be calculated if there are, respectively, less than 3 and 4 classes of frequency with 5 events (ANSCOMBE,1949; SCOSSIROLI et al., 1974), the fitting to the Poisson distribution was calculable in 59% of the cases and to

Table 2. Spatial distribution of larval nests per cluster observed in different years, vineyards and training systems; ns.,non significantly different with anova test; *, significantly different p < 0.05; **, significantly differentp < 0.01 I

Number of sampled vineyards

Differences between rows

Sylvoz 1980 1981 1982

1 15 12

n.s.

Pergola 1985 1986

7 1

Training system and year

I I

Shoot Basal

Distal

Basal

Cluster Distal

0.44 0.50 0.20

**

n.s.

0.32 0.20 0.05

0.38 0.34 0.10

0.38 0.35 0.13

n.s. n.s. n.s.

ns. n.s.

0.31 0.33

0.37 0.16

n.s.

0.33 0.26

0.35 0.23

n.s. n.s.

n.s.

** **

*

F. Pavan, V. Girolami and G. Sacilotto

364

Table3. Number of samples with Gariance less the mean and number ofsamplesfitting negatiue binomial distributions at different density ranges of grape berry moths Number of samples

Number of samples with S2 5 R

< 0.5 0.5-1.0 > 1.0

58 25 11

17 7 0

20 25 11

13 12

Total

94

24

56

Density range (larval nests per cluster)

Poisson Fitting Calculable n' n' %

the negative binomial distribution only in 28% of the samples (table 3). The Poisson distribution was not rejected in 48% of calculable cases and the negative binomial in 62% of calculable cases. The negative binomial distribution was also rejected in the 24 cases with variance less than the mean (table 3). At a density range below 0.5 larval nests per cluster the Poisson distribution is clearly more appropriate than the negative binomial distribution, while at a density range above 1.O larval nests per cluster the latter is preferable (table 3). The log mean and log variance of each data set are shown in fig. 2, with an indication of the distributions that each one fits best. Data sets fitting the Poisson were very close to the line of equal mean and variance, which by definition is the Poisson distribution, and normally have a low mean. Data sets fitting the negative binomial were more distant from the line of equal mean and variance, and means were higher. The fitting to different

/

0.5

Negative binomial Calculable Fitting n' 'n %

1

65 48 9

2 13 11

0 8 8

0 62 73

26

48

26

16

62

frequency distributions in relation to population density is well established (TAYLOR,1984). On the basis of a specific frequency distribution the relationship between larval nests per cluster (X) and percentage of infested cluster (1%)can be determined, and MOZZI (1983) as already reported by GIROLAMI for grapevine leaves infested by Panonychus ulmi (Koch). The probability of larvae-free clusters (Po)is given by: p0 -- e-" (eqn 1) on the basis of the Poisson distribution, and Po = 1 /[1

+ (it/k)k]

(eqn 2)

on the basis of the negative binomial distribution, where k = xZ/(Sz-')(ANSCOMBE,1949). is given by The percentage of infested clusters (I%) 1% = (]-Po)

. 100

(eqn 3)

and from eqns 1, 2 and 3 the following equations can be obtained:

I%

1

( x 2 test) the Poisson and the

= (1 -e-').

100 3 it

=

-In [ 1 -(I%/ loo)] (eqns4 and 5)

for the Poisson distribution and

h

a 0 C

.-a

3

0

Y

-F

Poisson A

-0.5

Negative Binomial Either Poisson or Negative Binomial

-1 -1

-0.5

0

0.5

1

log (mean)

Fig.2. Log mean-log variance of each larval nest per cluster plotted in comparison to line of equal mean and variance and with the indication of frequency distribution that each onefits. Diagonal line is a perfect fit to Poisson distribution

for the negative binomial distribution. The data sets for percentage infested clusters (1%) and larval nests per cluster (x)are close to the curve that links I% and on the basis of the Poisson distribution (eqn5) when I% is below 40% of infested clusters (0.5 1arval.nests per cluster) (fig. 3). Above this percentage the larval nests per cluster seems to be slightly underestimated by the Poisson and more closely described by the negative binomial distribution (table 4).

3.3 Weight loss estimate method The percentage weight loss (L)can be calculated on the bases of the average weight loss in damaged berries ( W), number of damaged berries per cluster ( A ) and average weight of cluster (C):

Second generation of grape berry moths

365

~

x =

-

In (1-1d100)

I

Substituting eqn 5 for X in eqn 9 links the percentage weight loss ( L )with the percentage of infested clusters ( I % ) on the basis of the Poisson distribution,

3.4 Empirical application of weight loss estimate method

To evaluate how well our new method estimated weight loss, we considered the precision of estimate of each sample and the variability of estimate in different cultivars and year. 3.4.1 Precision of estimate

The precision of estimate of the three parameters B, W and Cis reported in table 5. The 95% confidential interVal (C.I.950,Jwas calculated (standard error (SE) per 0 20 40 60 80 O0 Student’s t5%value). For practical purposes it is more important to know % infested clusters the C.I.950hof component ( W B / C ) 100 of eqn9 and Fig. 3. The percentage of infested clusters (1%) and the 10. eqn number of larval nests per cluster (X) for each sample in If the SEW,SE,, and SEc is known, the SE of ( W B / C ) comparison to the Poisson distribution for I% and R equation (SE,) is given by

’

Table 4. Percentage of infested clusters calculated, for different levels of larval nests per cluster, on the basis of the Poisson (eqn 4 ) and the negative binomial (eqn 6 ) distributions Negative binomial % infested Larval nests per Variance clusters (P) k (Z%) cluster (1) 0.1 0.2 0.3 0.4 0.5 0.6 0.7

9.8 0.9 1.o 1.5 2.0 2.5 3.0

0.10 0.21 0.33 0.46 0.58 0.71 0.84 0.98 1.11 1.25 1.94 2.66 3.40 4.14

L = ( W A / C ) * 100.

14.43 2.70 2.60 2.75 2.95 3.16 3.18 3.61 3.83 4.05 5.11 6.05 6.98 7.86

9.5 17.5 24.7 31.2 37.0 42.3 46.9 51.4 55.4 59.1 73.2 82.2 88.2 92.1

,,r

+ (SEB- +(SEc-6f)’ 6B

6C

Poisson % infested

clusters (I%) 9.5 18.1 25.9 33.0 39.3 45.1 50.3 55.1 59.3 63.0 71.1 86.5 91.8 95.0

(eqn 7)

The number of damaged berries per cluster ( A ) is obtained by multiplying the average number of damaged berries (tunnelled, and untunnelled rotten because of contact with tunnelled ones) per larva (B) and the number of larval nests per cluster (a): (eqn 8) and from eqns7 and 8 the following equation can be obtained ,

A = Bji

L = (WBjt / C ) . 100.

SEP=J(SEW-6f)’ 6W

(eqn 9)

The C.I.950hof component (W B / C) . 100 of eqns 9 and 10 is C.l.95yo= f SEf.t,, 100 and then, because Student’s t5%is about 2 for a number of more than 100 degrees of freedom, C.I.95%= & SEf.200. On the basis of the five considered cases the precision of estimation of ( W B / C ) .lo0 can be considered satisfactory. The extreme values calculated on the basis of C.1.95y0 are lower than the half of the mean only in one case (Verduzzo 83) and never higher than twice the mean. For practical purposes an error of this amount in weight loss estimate at harvest time is, in any case, more precise than that obtained with the method based on weighing the yield (or the cluster) as reported in the introduction.

3.4.2 Variability of weight loss estimate in different cultivars and years On the basis of eqns9 and 10 and data collected in 12 northeastern Italian vineyards, the percentage weight loss ( L )was estimated at the harvest time (table 6 ) . The percentage weight loss ( L ) varied in relation to the number of larval nests per cluster X and the values of parameters. On average L increased with the increase

F. Pavan. V. Girolami and G. Sacilotto

366

Table 5. Estimate of parameters B, W and C j o r dij@rent culticars and years on furnz A (see table 6 )

Cultivar and year

No. of damaged berries per larva ( B ) n Mean C.I.

Weight loss (8) in damaged berries ( W) n Mean C.I.

Merlot 83 Verduzzo 83 Tocai f. 83 Chardonnay 83 Chardonnay 84

33 28 42 62 74

4 4 4 4 4

n

= number

3.9 3.7 5.0 3.0 6.0

f 0.15

+ 0.12 '

k 0.42 i 0.13 f 0.19

0.94 1.02 1.10 1.03 0.76

Average weight (g) of cluster (C) n Mean C.1.

w k 0.07 w & 0.10 w 0.08 w k 0.06 w k 0.03

*

100 100 100 100 100

171 218 195 160 154

(WBIC). 100 Mean C.I. 2.14 k 0.69 1.73 0.91 2.82 f 1.18 1.93 2 0.51 2.96 k 0.64

c k3 ck6 ck 5 ck3

+

c+4

of sampling values and C.I. = 95% confidential interval

Table 6. Estimate of percentage weight loss (L) caused by the second generation of grape berry moths on the basis of empirical data (mean f S . E . ) , collected in two farms ( A and B ) , for different cultivars and years Cultivar and farm

No. of clusters

CkSE (g)

B+SE

R

I%

Lx

LP

Ltm

~~

Merlot A 79 A 82 B 83

96 48 I00

6.6 3.9 3.9 k 0.29

0.99 f 0.13 0.98 0.05 0.94 k 0.13

+

151 5 164k 8 171 f 6

3.51 0.54 0.33

96% 35% 26%

15.2% 1.2% 0.7%

13.9% 1.O% 0.5%

16.7% 1.1% 0.7%

Verduzzo B 83

100

3.7 f 0.23

1.02 f 0.17

218f 11

0.28

21%

0.5%

0.4%

0.4%

T0caif.A 79 A 80 A 81 A 82 B 83

I00 100 96 160 100

4.6 3.1 2.8 2.7 5.0 f 0.8

0.64 0.09 0.64 f 0.12 0.69 f 0.12 0.60 f 0.08 1.10 f 0.14

197f 7 173 k 7 1 2 6 k 10 140 f 6 195 k 9

3.38 0.70 0.92 0.17 0.42

95% 48% 56% 17% 3 1Yo

5.1 yo 0.8% 1.4% 0.2% 1.2%

4.5% 0.7% 1.3% 0.2% 1.O%

5.2% 0.8% 1.4% 0.2% 1.1%

Chardonnay A 81 B 83 B 84

96 100 100

3.3 3.0 k 0.24 6.0 f 0.36

0.96 f 0.09 1.03 f 0.10 0.76 f 0.06

116f 5 160 6 154k 7

1.61 0.62 0.74

71% 41 yo 49%

4.4% 1.2% 2.2%

3.4% 1.0% 2.0%

3.8% 1.1% 2.2%

4.0

0.86

164

1.10

49 ?o'

2.8%

2.5%

2.8%

Mean

+

+

Three estimates of L are reported: the first (L,) refers to eqn9 and of the samples; the second ( Lp )refers to eqn 10; the third (Lbn)refers to eqn 9 and R calculated for points from I % on the basis of Taylor's equation log (S,) = 0.096 + 1.093 log X and inverse equation (eqn 6) ( B = number of damaged berries per larva; W = average weight loss in damaged berries; C = average weight of the cluster; X = larval nests per cluster I% = percentage of infected clusters)

of it, but for the similar values of f the damage is different in different years and cultivars (table 6). The number of damaged berries per larva (B), the average weight loss in damaged berries (w)and the weight of the clusters ( C ) varied in relation to cultivar and year. The differences in B are due mostly to the different spread of rots from tunnelled berries (directly damaged by larvae) to untunnelled berries in contact (indirectly damaged); the number of these latter cases can vary in relation to cultivars, meteorological conditions and anti-grey mould treatments (VALLI, 1980; PAVAN et al., 1989a, 1993a; PAVAN and SBRISSA,1994). The differences in W are due mostly to the different levels of dehydration of the berries induced by dry or humid condition during August and due to the harvest period of cultivars. Dehydration is higher on late-harvested cultivars since the injured berries have more time

to dehydrate (PAVAN et al., 1993a; PAVAN and SBRISSA, 1994). The wide variability of B, Wand C affected the percentage weight loss (L), but this was not more than twice or less than half the mean for the same percentage infested clusters. For example, at 50% of infested clusters (I%) the percentage weight loss ( L ) would vary from 0.8% to 3.0% with a mean of 1.5%. A weight compensation by undamaged berries surrounding damaged ones has been observed for late-harvested cultivars (PAVAN and SBRISSA,1997). Comparing L, (percentage weight loss estimated on the basis of eqn9 and j t of the samples) and Lp (percentage weight loss estimated on the basis of eqn 10) revealed that L, slightly underestimates the damage because eqn 10 refers to the Poisson distribution. The negative binomial distribution (Lbn)gives a most precise

Second generation of grape berry moths

367

estimate of percentage weight loss, but the impossibility of obtaining X with analytical methods from the inverse of eqn 6 leads to the use of eqn 10 instead. This usually estimates about 10% lower losses. The percentage weight loss per larva per 100 clusters, estimated in northeastern Italy, was on average five times lower than that estimated on the basis of weighing the yield (or clusters) in Germany (0.02% instead of 0.1 Yo)(KASTand MUNDER,1990).

3.5 Economic injury level The method proposed for estimating weight loss can be used to establish the economic injury level (EIL). Insecticide treatments against the second generation of the grape berry moths are worthwhile if the cost of control (C,) is less than the economic injury (EI): C, < EI.

the basis of experience (e.g. cultivar, climatic conditions). On the basis of eqn 15 and data collected in northeastern Italian vineyards, the percentage of infested clusters corresponding to the economic injury level (EIL) was estimated (PAVANand GIROLAMI,1986; PAVANand SBRISSA, 1997).The economic injury levels for the second generation of grape berry moths can vary between 10% and 30% for early-harvested cultivars and between 40% and 50% for late-harvested cultivars, where the possibility of a weight compensation is considered. For early-harvested cultivars with high susceptibility to rot, the economic injury level must not exceed 20% of infested clusters to avoid qualitative damage of musts (ROHERICHand SCHMID,1979; PAVANand GIROLAMI, 1986).

(eqn 11)

In this case a 100% effectiveness of the control is assumed. The EI for the second generation of grape berry moths is a function of the percentage weight loss (L), the yield in tons (t) per hectare (Y) and the price per ton of wine grapes (P),

4 Conclusion

The position of clusters within vine can influence the number of larval nests per clusters. Therefore, a valid estimate of larval nests per cluster is not possible choosing the clusters at random. It is necessary to use an established sampling scheme. The larvae on clusters are almost randomly disEI = L YP/lOO. (eqn 12) tributed below an average of 0.5 larval nests per cluster, but slightly aggregated above 0.5 larval nests per cluster. From e qns l l and 12 the following equation can be Therefore, below 0.5 larval nests per cluster (about 40% of infested clusters), the relationship between perobtained: and larval nests per cluster centage infested clusters (I%) c,< L YP/lOO (eqn 13) (x)on the basis of Poisson distribution is useful for and then from eqns 10 and 13 the percentage weight loss estimating X when I%is known. Above 0.5 larval nest can be obtained above which treatment is worthwhile per cluster (X), the Poisson distribution slightly underestimates X. (L'): The estimate of the percentage weight loss (L), for the similar larval infestation varies with the cultivar and L'> C,.lOO/(YP) year, but the risk of error can be reduced by considering = ( W B / C ) {-In [ 1 -(I% / loo)]} 100 (eqn 14) matters under dry meteorological conditions and on In eqn 14 the percentage of infested clusters (I%) cor- late-harvested cultivars, a lower number of damaged responding to the economic injury level (EIL) can be berries per larva (B)and a higher average weight loss of damaged berries ( W ). obtained as follows: The weight loss estimated by this new method can be also used to establish the economic injury level for the [C,C / (Y P W B ) ]= -In [ 1 - (EIL / loo)] second generation of grape berry moths. Action thre* -[C,C/(YPWB)] = ln[l-(EIL/100)] sholds are used because control measures must be taken before the economic injury level is exceeded. For this 3 exp[-C, C / ( Y P W B ) ]= 1 -(EIL/ 100) purpose, action thresholds based on minimal amounts EIL/ 100 = 1 -exp [- C, C / (Y P WB)] of male catches with pheromone traps (BOLLERand REMUND,1981; SCHRUFT,1989) were proposed in 3 EIL = 100 { 1 - exp [- C, C / (Y P WB)]} different European grape-growing areas. In north(eqn 15) eastern Italy, action thresholds were proposed to be based on larval tunnels observed 10-1 5 days after the Because the treatments against the second generation emergence peak and curative treatments (PAVANand of grape berry moths must be carried out in July when GIROLAMI,1986; PAVANet al., 1987, 1989b, 1993b; many parameters are unknown, the economic injury PAVAN and SBRISSA,1997). level is only a provisional threshold. In fact, only the cost of treatment (C,) is known, while the other parameters can only estimated. The yield (Y), the price of Acknowledgements grapes (P) and the average weight of the cluster (C)can We wish to thank A. H. PURCELLof University of California be forecast with a good precision; the weight loss in for improvements in the manuscript made possible by his damaged berries (W)and particularly the number of comments. We thank the owners of the vineyards where the damaged berries per larva ( B )can be only supposed on research was carried out.

368

F. Pavan. V. Girolami and G. Sacilotto

larval sampling and treatment period for grape moths. In: Integrated pest control in viticulture. Ed. by Cavalloro, ANSCOMBE, F. J., 1949: The statistical analysis of insect counts R. Rotterdam: Balkema, 39-49. based on the negative binomial distribution. Biometrics -; GIROLAMI, V.; CECCHINI, A,; TURBIAN, E., 1993a: Evo5, 162-173. luzione dei danni delle tignole dell’uva, Lobesia hotrarzu BLISS,C. I.; FISHER, R. A,, 1953: Fitting the negative binomial (Den. e Schiff.) ed Eupoecilia ambiguella (Hb.), nell’Italia distribution to biological data. Note on the efficient fitting nord-orientale e lotta insetticida. Redia 76(2), 417431. of the negative binomial. Biometrics 9, 176-200. G.; CECCHINI, A,, 1993b: Voli degli adulti, -;-; SACILOTTO, U., 1981: Relations entre les captures BOLLER, E.; REMUND, dinamica delle popolazioni larvali e nuovi criteri di conde Eupoecilia arnbiguella et le pourcentage de grappes trollo delle tignole della vite, Lobesiu botrana (Den. e attaquees. Boll. Zool. agr. Bachic. (11) 16, 21-22. Schiff.) ed Eupoecilia ambiguella (Hb.). Frust. Ent. (ns) GIROLAMI, V., 1981: Evaluation des degits dus aux vers de la 16, 79-88. grape. Boll. Zool. agr. Bachic. (11) 16, 16-18. ROEHRICH, R. 1978: Temoignage d’experimentateurs: Les tor-; MOZZI,A,, 1983: Distribution, economic threshold and deuses de la grappe (recherches sur la nuisibilitk de Eupoesampling methods of Panonychus ulmi (koch). In: Statcilia ambiguella Hb. et Lobesia botrana Den. et Schiff.). istical and mathematical methods in population dynamics Defense Vegetaux 191, 106-125. and pest control. Ed. by CAVALLORO, R. Rotterdam: -; SCHMID, A., 1979: Lutte intigree en viticulture. Tordeuses Balkema, 90-101. de la grappe: evaluation du risque, determination des Y., 1974: Some generalisation of the T-method HOCHBERG, periodes d’intervention et recherche de methodes de lutte in simultaneous inference. J. Multivar. Anal. 4, 224234. biologique. In: Internationales Symposium der IOBCKAST, W. K.; MUNDER,H., 1990: Untersuchungen zur WPRS iiber Integrierter Pflanzenschutz in der Land-und Befalls-Verlust-Relation und wirtschaftlichen SchaForstwirtschaft. Wien: Gistel Druck, 245-254. densschwelle beim Einbindigen Traubenwickler (Eupoe- SACILOTTO,G., 1981: Valutazione dei danni di Lobesia cilia ambiguella Hbn.). Z. Pfl. Krank Pfl. Schutz 97(1), botrana Schiff. Opportunita d’intervento e conseguenze 76-83. dei trattamenti fitosanitari. Tesi di laurea, Universita di F. M.; GONCALVES, M. L.; SOBREIRO, J. B.; LAVMARTINS, Padova, Italy. ADINHO,A. M. P., 1989: Control of Lobesia botrana SCHRUFT, G., 1989: Forecasting and economic threshold of Schiff. in Oeste (Portugal): influence of different treatthe European grape berry moth (Eupoecilia ambiguella ments against the second and the third generations. In: Hbn.) in southern west-Germany vineyards. In: Plant proPlant protection problems and prospects of integrated tection problems and prospects of integrated control in control. in viticulture. Ed. by CAVALLORO, R. Brussels: viticulture. Ed. by CAVALLORO, R. Brussels: ECSCECSC-EEC-EAEC, 101-1 08. EEC-EAEC, 681-686. PAVAN,F.; GIROLAMI, V., 1986: Lotta guidata alle tignole -; WOHLFARTH, P.; WEGNER,G., 1988: Untersuchungen zur della vite su uve da vino nell’Italia nord-orientale. L’InNotwendigkeit der chemischen Bekampfung des Einform. Agrario 42 (30), 3541. bindigen Traubenwicklers (Eupoecilia ambiguella Hbn.) -; SBRISSA,F., 1994: Dannosita delle tignole della vite, im Hinblick auf ein Schadschwellen-Konzept. Die WeinLobesia botrana (Den. e Schiff.) ed Eupoecilia ambiguella Wissensch. 43, 174-185. (Hb.), su cultivar a maturazione tardiva nell’Italia nord- SCOSSIROLI, R. E.; CLEMENTEL, S.; SCOSSIROLI, S., 1974: orientale. Frust. Ent. (ns) 17,43-53. Metodi statistici per I’ecologia. Analisi delle distribuzioni di frequenza. Zanichelli, Bologna, Italy. -;-, 1997: Perdite in peso causate dalla seconda generazione delle tignole della vite, Lobesia botrana (Den. e Schiff.) ed TAYLOR,L. R., 1961: Aggregation, variance and the mean. Eupoecilia ambiguella (Hb.), e soglie d’intervento su uve Nature 189,732-735. da vino. Frust. Ent. (ns) 20, (in press). -, 1984: Assessing and interpreting the spatial distributions -; DALLAMONTA,L.; TURBIAN, E., 1989a: Influence des of insects populations. Ann. Rev. Ent. 29, 321-357. divers anticriptogamiques sur les degits des vers de la VALLI,G., 1975: Lotta integrata nei vigneti. Ricerche e valugrappe. In: Plant protection problems and prospects of tazioni preliminari sulle Tignole. Notiz. malatt. piante integrated control in viticulture. Ed. by CAVALLORO, R. 92-93,4074 19. Brussels: ECSC-EEC-EAEC, 1 19-128. -, 1980: Studio sulla correlazione tra tignole della vite e -; Duso, C.; PAVANETTO, E., 1989b: Integrated pest control muffa grigia. In: 2” incontro sulla difesa della vite. Ed. by in viticulture in north-eastern Italy. 2. Grape moths and Regione Lazio, 21-25. leafhoppers. In: Plant protection problems and prospects of integrated control in viticulture. Ed. by CAVALLORO, Author’s address: F. PAVAN (corresponding author), Dipartimento di Biologia applicata alla Difesa delle Piante, UniR. Brussels: ECSC-EEC-EAEC, 671-679. -; SACILOTTO, G.; GIROLAMI, V., 1987: Damage evolution, versita degli Studi, via delle Scienze 208, 33100 Udine, Italy

References