Denson Kelly McLain and Stephen Donald Mallard.
Sources and adaptive consequences of egg size variation in Nezara viridula (Hemiptera: Pentatomidae).
Psyche 98:135-164, 1991.

Full text (searchable PDF, 2200K)
Durable link: http://psyche.entclub.org/98/98-135.html

The following unprocessed text is extracted from the PDF file, and is likely to be both incomplete and full of errors. Please consult the PDF file for the complete article.

SOURCES AND ADAPTIVE CONSEQUENCES OF EGG SIZE VARIATION IN NEZARA VIRID ULA
(HEMIPTERA: PENTATOMIDAE)
BY DENSON KELLY MCLAIN* AND STEPHEN DONALD MALLARD Department of Biology
Landrum Box 8042
Georgia Southern University
Statesboro, GA 30460
Very little is known about the relationship in insects between female size and egg size (e.g. Price and Wilson, 1976; Wiklund and Karlsson, 1988) or how that relationship responds to selection pressures imposed by changing ecological and physiological envi- ronments (e.g. Moore and Singer, 1987). A few studies have, how- ever, documented some of the fitness consequences of variation in egg size. For example, offspring from larger eggs may hatch more successfully (Richards and Meyers, 1980), develop more rapidly (Steinwascher, 1984), or attain larger size (Johnson, 1982). Assuming that egg size is positively correlated with offspring fitness, an optimal egg and clutch size may evolve (Smith and Fretwell, 1974). The optimum may depend on the degree and nature of larval competition (Parker and Begon, 1986; Ives, 1989), relative safety of egg versus larval stages (Shine, 1989; Nassbaum and Schultz, 1989), or variation in quality of larval habitats (Cap- inera, 1979; McGinley et al., 1987). In addition, the optimum may vary from female to female within a population as a consequence of differences in body size and foraging efficiency (Parker and Begon, 1986; Shine 1988).
The present study documents egg size variation in the southern green stink bug, Nezara viridula L. (Hemiptera: Pentatomidae) and examines the life history consequences of such variation. Female size, mate size, egg mass number (i.e. first, fifth, tenth, or fifteenth mass laid), egg mass size, and location within the egg mass (inner *TO whom correspondence should be addressed. Manuscript received 6 May 1991
135
Pnrhe 98:135-164 119tl[) http:ffp~yche.cnlclub.or<>W8Ìö8-13 html



================================================================================

136 Psyche [vo~. 98
versus outer egg rows) are assessed for their contribution to varia- tion in egg size. We also examine the correlation between egg size and offspring hatching success, ultimate adult size, probability of survival to adult stage, and development rate. Collection and rearing.
Southern green stink bugs, Nezara viridula, were reared in the laboratory from the eggs of 30 females collected in mid-April (1988 and 1990) from wild mustard growing in an abandoned pas- ture on SR 24, 6 miles north of its junction with US 80 in States- boro, GA (USA). Each female was housed individually in a 1-liter jar supplied with 2 pole beans that served as both a food and water source. A strip of white paper towel was suspended from the rim of each jar as an oviposition substrate. Eggs were hatched in petri dishes containing pole bean onto which nymphs moved 1-2 d after hatching. Nymphs, in groups of 40, were reared in 4-liter cartons supplied with fresh pole beans and shelled, raw peanuts. Beans and peanuts were replaced every 5 d. Insects were maintained on a 14L: 10D photophase at 20-23OC and 80-85% RH. Cartons with fifth instar nymphs were checked daily for adults which were then sequestered by sex. Because adults do not mate until they are over 1 week old (Harris and Todd, 1980), there was no opportunity for females to mate within rearing cartons. Mating and maintenance of females.
One male and 1 female, each 10-d-old virgin adults, were placed in 1-liter jars that served as mating chambers. In all cases, the males and females that were housed together were derived from different mothers. Otherwise, individuals were assigned to mating chambers randomly and without regard to size. Each mat- ing chamber was stocked with 2 pole beans. Once a pair had termi- nated copulation, the female was removed from the mating chamber to a 112-liter carton (oviposition carton). Male and female size (width across the pronotum at the humeral angles) was mea- sured to the nearest 0.125 mm with an ocular micrometer on a dis- secting scope.
Females were allotted 1 month in which to initiate copulation. Mating activity was checked 1, 6, and 12 h into photophase. Since



================================================================================

199 11 McLain & Mallard 137
the duration of copulation averages 30-40 h (Harris and Todd, 1980) it is unlikely that many, if any, matings went undetected. No egg masses were found in mating chambers although over 60% of oviposition occurs within 1 d of the termination of copulation (Harris and Todd, 1980).
Once-mated females remained isolated in their cartons for the remainder of their lives. Cartons were restocked with a fresh pole bean every 5 d to prevent foodlwater supplies from drying out. White paper towel strips were suspended from carton rims as oviposition substrates. Cartons were checked daily for egg masses. Egg size.
Egg diameter (= size) was measured to the nearest 0.01 mm with an ocular micrometer on a compound microscope at a magni- fication of 100X. In 1988, eggs were measured from the first, fifth, tenth, and fifteenth (if available) masses laid. Eighty females were included in this analysis of which 32 survived to lay 15 egg masses. Five eggs were measured from each mass, two from the middle row of the mass, two from one of the outer rows, and one of unspecified location. Egg masses were typically hexagonal arrays of 50-100 eggs in 5-10 rows.
Population density and egg size.
An additional 72 females were permitted to mate under condi- tions of variable density (1 male + 1-7 females). Upon initiation of copulation, the pair was removed and the female treated as out- lined above. We measured the size of the female and the sizes of five eggs from the first egg mass laid by each female. Egg size and offspring fitness.
In 1990, 5-10 outer row and 5-10 inner row eggs were taken from the first egg mass of 20 experimental females. Eggs were separated by hand and placed singly into 1-liter jars supplied with two pole beans. Hatch rate was 83.6%. Beans were replaced every 5 d (=benign conditions). This rearing regime resulted in 29.3% mortality for nymphs that hatched. The egg size, time (d) from egg hatch to adulthood, and pronotal width were recorded for each individual surviving to adulthood. Egg and adult size were mea- sured as before.




================================================================================

138 Psyche [vo~. 98
Also, 5-10 inner and outer row eggs were separated from each of 15 initial egg masses and placed singly into 1-liter jars supplied with a pole bean. The pole bean in each jar was replaced every 7-9 d (=harsh rearing condition). This rearing regime resulted in 41.0% mortality. The egg size and survival of hatched nymphs to adulthood were recorded for each case.
The percent egg hatch depends upon humidity (unpublished observation). Thus, to determine the relationship between egg size and hatch rate under humidity-restricted conditions, some egg masses were incubated in petri dishes that did not contain pole bean (In the presence of beans in the petri dish the hatch rate is approximately 100% for fertilized eggs). Forty masses (first laid by each of 40 females) were developed under the humidity- restricted condition. Two unhatched eggs, one each from inner and outer rows were measured. An equal number of hatched eggs, posi- tioned adjacent to unhatched eggs, were measured. The following scheme was employed to select eggs for measurement without bias: we chose the first unhatched egg and adjacent hatched egg that were at least 4 eggs in from the left margin of the row. Devel- opment without a bean resulted in less than 50% hatch for fertil- ized eggs (Fertilization is evidenced by a change in egg color from cream to orange). A paired-difference t-test was used to determine if mean size differed between hatched and unhatched eggs. Statistical methods.
For some analyses, egg sizes were converted to relative sizes (i.e., yij/y., where yij is the size of egg i of female j and ymj is the mean egg size for female j). Thus, variation in egg size associated with placement in the mass or associated with egg mass number e.g. first versus tenth mass laid) is not confounded with variation between females in egg size.
The joint effects of independent variables on egg size were tested with multivariate general linear hypotheses of the form: egg size = constant + TI + T2 + ... + Tn + El + E2 + ... + En, where Ti is a category variable (e.g., egg mass number) and Ei is a continu- ous variable (e.g. female size). Statistical analyses were performed on a microcomputer using the Systat system for statistics (Wilkin- son, 1987). In linear models, we report only variables contributing significantly to explained variation (= R~). Absence of interaction



================================================================================

19911 McLain & Mallard 139
terms connotes an absence of significance. Interaction between variance sources I and J is indicated as (I)*(J) in models below. For univariate models, significance of the Pearson's correlation coefficient, r, is tested with a t-test (Edwards, 1985). Since multiple eggs were measured or hatched for each female (mother) in 1988 and 1990, there is potential for degrees of free- dom to be inflated by nonindependent data points. Therefore, all ANOVA tests and ANCOVA models include "mother" and its interaction terms as sources of variation to group data within females. This reduces the error degrees of freedom and increases the error mean square. The effects of "mother" and its interaction terms are reported only when significant (Pc0.05). For mutivariate analyses incorporating egg mass number as a source of variation, only data from females laying at least 15 egg masses are used. This reduces the sample size from 80 to 32 moth- ers. Multivariate analyses of the effect of egg size on components of progeny fitness required at least four observations per sex of offspring per mother. This reduces the sample sizes to 10 or 12 mothers for tests on, respectively, females or male progeny. Variation between females.
Egg size exhibits a relatively high degree of variation; the coef- ficient of variation in egg size (CV=5.75 reflecting within- and between-female variation) was 98% as large as that for female size (CV=5.89). Models incorporating male size, female size, the inter- action between male and female size, or total number of egg masses (<I5 or >IS), either singly or in any combination, failed to explain a significant proportion of the variation between females in CV in egg size (For all models: ~~e0.02; P>0.40; N=80). Differences between females in egg size accounted for 57.2% of the total observed variation in egg size (F= 17.15; df=79,1012; P<0.001). Female size accounted for 35.8% of the variation between females in lifetime mean egg size (F=35.91; df=l,76; Pe0.001; Fig. 1). Mate (male) size did not contribute to the expla- nation of variation in mean egg size across all masses (F=2.23; df=l,74; P=0.14 when incorporated with female size in a linear model). Male size did explain a significant proportion of the between-female variation in egg size for the first mass laid after



================================================================================

Psyche
82 -
79 -
76 -
73 -
0
70 -
0
15 masses
67
F l
. ~ . ~ ~ I ~ i m i n l ~
[Vol. 98
0
Female size
Figure 1. Correlation between female size (width [mm] at humeral angles) and lifetime mean egg size (diameter [0.01 mm units]). copulation, but did not correlate with egg size in later masses (Table 1 ; Fig. 2).
When population density (no. bugslcontainer) and parent sizes are included in the analysis of mean egg size (first mass laid), the best model is mean = constant + density + (female size)*(male size) (F=24.15; df=2,71; ~~~0.42; P<0.001). Here, population den- sity alone accounts for 25.7% of the variation (F=24.21; df=l,72; P<0.001) while the interaction between parent sizes contributes another 15.8% increment in explained variation in mean egg size (F=17.36; df=l,71; P<0.001 for the increment). Variation within females.
Within females there was a relatively high degree of variation in egg size both across a lifetime (mean CV=4.44; SD=3.59; range 0.64-26.29; Nhies=80; Neggs=1360; year=1988) and in a single



================================================================================

19911 McLain & Mallard 141
Table 1. Effect of male (mate) and female size on mean egg size as a function of egg mass number in the sequence laid. Male + female; multiple regression (d.f.=2,75). Male (d.f.=1,78), Female (d.f.=1,76): univariate regression. Increment (d.f.=1,75): increase in explained variation due to addition of male size to regres- sion already including female size.
Female + Male Female Male Increment
Mass 1:
Mass 5:
Mass 10:
F
R~
P
Mass 15:
egg mass (mean CV=5.29; SDw=1.31; range 2.92-8.04; Nfemales=20; Neges=236; year=1990). Across a lifetime, the CV for egg size did not differ as a function of egg mass number (Kruskal- Wallis test for rank in CV; x2=4.09; df=3;P>0.05). Multivariate analysis of within-female variation in relative egg size for females laying 15 masses revealed significant effects for placement within the egg mass (F=6.59; df=l,256; R~=o.o~) and egg mass number (in the sequence laid) (F=16.82; df=3,256; R~=o.~o) (Table 2). Neither "mother" nor any of the interaction terms was significant (Fc0.02; ~~~0.002; P>0.05). The mean size of eggs in inner rows was larger than that of eggs in outer rows (In 1988: paired difference t-test, t=3.89; N=80; P<0.001 for all egg masses. In 1990: paired difference t-test, t=3.78, N=19, P<0.01 for first egg mass laid) (Fig. 3). The mean size of the inner row eggs was significantly positively correlated



================================================================================

142 Psyche [Vol. 98
82 : Mass number 1
Mass number 5
82 : å ft
80 :
80 :
0
78 :
76 :
å ´ * å´å´ :* :
74 :
:
72 :
å
å
Mass number 10
, , . ,;. , , , , , ,*, , , . . . . , , . , . , 68
6.0 6.5 7.0 7.5 8.0 8.5
Male size
Mass number 15
68
6.0 6.5 7.0 7.5 8.0 8.5
' Male size
Figure 2. Correlation between male size (width [mm] at humeral angles) and egg size (diameter [0.01 mm units]) as a function of egg mass number (in the sequence laid).
with the mean size of outer row eggs (Table 2). Thus, females lay- ing larger inner row eggs also laid larger outer row eggs relative to other females. Within a row, egg sizes are correlated; the sizes of the tw'o measured inner row eggs were highly correlated (r=0.85; t=29.24; P<0.001; N=273) as were the sizes of the two measured outer row eggs (r=0.79; t=22.97; P<0.001; N=273). Relative egg size was larger in later than earlier egg masses (Table 2; Fig. 3). Also, egg mass size (number of eggs) decreased with egg mass number (in the sequence laid) (Kruskal-Wallis test based on rank in mass size, x2=131.41; df=3; P<0.01; Table 2). However, relative egg mass size explained very little of variation in relative mean egg size (F=4.10; df=l,264; ~~~0.02; P=0.04; comparison across a lifetime). The preceding analysis is



================================================================================

19911 McLain & Mallard 143
Table 2. Relationships between egg mass number (in the sequence laid) and mass size (number of eggs), egg size, and intra-mass correlation (r) in egg size. In = inner egg row, Out = outer egg row. InIOut = correlation between mean size of inner and outer row eggs within the same egg mass. In/In and OutIOut = correlation between sizes of two measured eggs from, respectively, inner and outer rows of the same egg mass (P<0.05 for all correlations). N = number of egg masses sampled. SD for each mean is given in parentheses. Mass Size Egg size1 r
Mass N No. eggs Inner row Outer row InlOut In/In OutIOut 1 unit = 0.01 mm.
complicated by the fact that changes in physiological condition in later life may impact egg mass size erratically and much more than does egg size. Therefore, the influence of mass size on mean egg size was examined for the first mass laid using multiple regression models to control for the effect of female size. The following model explained the greatest amount of variation in mean egg size (F=11.24; df=2,74; ~~=0.23; P<0.001): mean egg size [range=69.5-80.51 = 75.39 - 0.22 (mass size [range=29-1171) + 0.027 (mass size)*(female size [range=6.38-9.001). Step wise regression retains both mass size and the interaction term. Thus, mass size and egg size are significantly negatively correlated. Progeny fitness and egg size.
Male and female progeny differed in adult size (female larger) and rate of development (female slower) (Table 3). Therefore, effects of egg size are treated separately for each sex. The sexes did not differ, however, with regard to egg size, although the inter- action (mother)*(sex) was significant (Table 3). Location within the egg mass did not differ by sex (Kruskal-Wallis test, x2=~.21; df=l; P~0.05).




================================================================================

Psyche
[Vol. 98
Ej Outer row
4 Inner row
I 5 I0 15
Egg mass number
Figure 3. Rank in egg size across a lifetime as a function of location within the egg mass (inner or outer egg row) and egg mass number (in the sequence laid). For both males and females, egg size within an egg mass was significantly positively correlated with adult size and significantly negatively correlated with development time (Table 4; Fig. 4 and 5). The effect of egg size on adult size differed by sex (i.e. inter- cept of the regression was higher in females; F=28.80; df=l ,I1 1; P=O.OOl in ANCOVA model: adult size = constant + egg size + mother + sex). Both egg size (Fz42.85; df=l,lll; Pc0.001) and mother (Fz2.88; df=9,111; P=0.004) had significant effects in this model. This may reflect that females have, on average, ,longer development time and, therefore, can achieve larger -adult size for a given egg size. In an ANCOVA model which al& incorporated the interaction between sex and egg size, the interaction was not significant (F- 1.3 1 ; df=l, 1 10; P>0.05), indicating homogeneity of slopes. Similarly, the intercept of the regression of egg size on



================================================================================

19911 McLain & Mallard 145
Table 3. ANOVA of effect of sex and mother on progeny traits: (A) adult size, (B) development time, and (C) egg size.
Source of Variation DF MS F P
A. Adult size: sex 1 32.60 81.69 <O.OOl
mother 17 0.54 1.14 0.326
sex*mother 17 0.33 0.83 0.656
error 12 1 0.40 - -
B. Development: sex 1 122.34 9.54 0.002
mother 17 16.02 1.25 0.237
sex*mother 17 13.02 1.02 0.447
error 121 12.82 - -
C. Egg size: sex 1 63.39 3.80 0.054
mother 17 42.65 2.56 0.002
sex*mother 17 29.52 1.77 0.040
error 121 16.69 - -
development time was higher in females (F=64.26; df= 1,111 ; P<O.OOl) while mother (F=5.08; df=g71 11; P<O.OOl) and egg size (F=58.99; df=l 11 1 ; Pc0.001) had significant effects. Among males, individuals derived from inner row eggs were larger as adults than those from outer row eggs (analysis based on relative measures7 Fz29.35; df=1756; P<O.OOl; R2=0.47; neither mother nor the interaction between location and mother were sig- nificant). Also, males derived from inner row eggs developed faster than those from outer row eggs (Fz98.79; df=1756; P<O.OOl; R2=0.34 for effect of location). Here, there were also significant effects for mother (Fz2.58; df=ll,56; P=O.Ol; R2=0.10) and the interaction (mother)*(location) (F= 10.00; df= 1 1,56; Pc0.001; R2=0.38). These location effects may reflect the greater average size of eggs in the mass interior (ANOVA of egg size: Location, Fz47.30; df=1,56; Pc0.001; R2=0.34. Interaction with mother7 Fz2.23; df=l l,56; P=0.025; R2=0.18). Sample sizes for females were insufficient to permit corresponding analyses. Relative adult size and development time were significantly negatively correlated in males (F= 19.07; df=1765; P<O.OOl; R2=0.22; effect of mother was nonsignificant; Fig. 6) and females



================================================================================

146 Psyche [vo~. 98
Table 4. ANOVA of effect of egg size and mother on (A) adult size and (B) development time in (1) males and (2) females. For each mother, all measures have been converted to relative values by dividing by the mean of the mother. The effect of relative measures is to de-emphasize between-mother effects due to differences in scale (Contrast effects of mother in Tables 4 and 5). Proportion of variation explained, R~, is: male size, 0.38; female size, 0.53; male development time, 0.68; female development time, 0.7 1.
Source of Variation
A. Adult size,
1. Males:
2. Females:
B. Time,
1. Males:
2. Females
egg size
mother
egg size*mother
error
egg size
mother
egg size*mother
error
egg size
mother
egg size*mother
error
egg size
mother
egg size*mother
error
(Fz52.70; df=1,40; P<O.OOl; R2=0.57; effect of mother was non- significant). In the best multivariate model predicting adult size of females (R2=0.70), time (Fzl4.55; df=1,38; P<O.OOl), egg size (F=ll.Ol; df=1,38; P=0.002), and their interaction (F=l3.ll; df=1,38; P=O.OOl) had significant effects while the effect of mother was nonsignificant (F=0.61; df=9,38; P>0.05). The best model for males (R~=o.~o) included mother and its interactions with development time and egg size. However, none of the individual sources of variation were significant (Fz1.39-1-89; df=l l,46; P>0.05).




================================================================================

McLain & Mallard