C. B. Montllor, Department of Environmental Science, Policy, and Management, 201 Wellman Hall, University of California, Berkeley, CA 94720-3112, U.S.A. E-mail: firstname.lastname@example.org
Abstract 1. Natural populations of pea aphids in California contain at least two facultative bacterial secondary symbionts (pea aphid secondary symbiont, PASS, or pea aphid rickettsia, PAR) in a range of frequencies throughout the state.
2. Two pea aphid clones without either of these facultative associates failed to reproduce in the first 8 days after the final moult if they had been heat-stressed for a period of about 4 h at 39 °C as 1-day-old larvae in the laboratory.
3. Aphids infected artificially with PASS, however, were able to produce up to 48% of the normal complement of offspring produced by PASS-positive aphids that had not been heat-stressed. Clones infected artificially with PAR did not have the same advantage as those with PASS after heat stress.
4. In aphids without PASS or PAR, heat stress reduced the number of bacteriocytes (in which the obligate primary symbiont, Buchnera, resides) to 7% of non-heat-stressed aphids, while aphids with only PASS retained 70% of their bacteriocytes. Bacteriocytes in aphids with PAR but not PASS were reduced to 42% of controls.
5. When larvae were heat-stressed as older instars (5 days old), a similar pattern emerged, though the effect of heat stress was less extreme. Clones containing PASS produced the most offspring, three to 14 times as many as aphids without PASS or PAR. Aphids with PAR only, or PASS and PAR together, had reduced or no advantage over aphids without facultative symbionts.
6. Aphids of all clones that had been heat-stressed as later instars gave birth to a variable number of stillborn offspring. Aphids without facultative symbionts produced the most stillborn larvae.
7. Field studies showed a higher incidence of PASS in aphids collected in California in summer compared with aphids from the same sites collected 2–4 months earlier. The difference was significant in two of three widely dispersed locations.
The biological importance of the primary symbiont Buchnera to the aphid host has received much attention, and many of the contributions that the symbiont makes to its host have been characterised at the physiological, ecological, and molecular levels (for reviews see Baumann et al., 1995; Douglas, 2000). The genetic map and recently sequenced Buchnera genome corroborate the obligate nature of this symbiosis for both host and symbiont (Charles & Ishikawa, 1999; Shigenobu et al., 2000). The widespread occurrence of facultative bacterial symbionts that are not obligate partners has been little noted as a potential factor in the biology of the pea aphid host, but is beginning to receive more attention (Chen et al., 1996, 2000; Chen & Purcell, 1997; van der Wilk et al., 1999; Fukatsu et al., 2000). Although the secondary symbiont of the pea aphid was described from ultrastructural studies almost 30 years ago (McLean & Houk, 1973), its effect on the aphid has been investigated only recently (Chen et al., 2000). This is the first study describing unequivocal benefits of harbouring such bacteria.
Two facultative symbionts of the pea aphid Acyrthosiphon pisum (Harris) designated pea aphid secondary symbiont, or PASS (a bacterium in the γ-3 subdivision of the Proteobacteria) (Unterman et al., 1989) and pea aphid rickettsia, or PAR (a rickettsia in the α subdivision of the Proteobacteria) (Chen et al., 1996), were detected in the majority of California pea aphid populations examined during 1995 (Chen & Purcell, 1997), 1999 (C. Montllor and A. Maxmen, unpublished), and 2000 (F. Barre, unpublished). They occurred in variable frequencies of up to 80% for PASS and 50% for PAR, however the effects of PASS or PAR ranged from neutral to pathogenic in laboratory studies, and no evidence for horizontal transmission has been found (Chen et al., 2000). The theory that the net benefit of these bacteria to the pea aphid must be greater than or equal to 1 (given no horizontal transmission) relative to uninfected individuals (Fine, 1975) suggests that both PASS and PAR confer a net selective advantage to infected pea aphids.
All pea aphids contain the primary bacterial symbiont Buchnera aphidicola (also γ-3 Proteobacteria), which is essential to the proper development and growth of the aphid host (Douglas, 1998a). With respect to PASS and PAR, aphids from natural populations in California had one of four possible combinations of these bacterial associates: they were PASS- and PAR-negative (P–R–), PASS-positive (P+), PAR-positive (R+), or were host to both bacteria (P+R+). Worldwide, pea aphids may harbour additional facultative symbiotic bacteria, including Spiroplasma described from Japanese aphids (Fukatsu et al., 2001) and three Enterobacteriacea designated T-type, U-type, and R-type from a variety of locations in the United States (Sandstrom et al., 2001). The R-type was virtually identical to what has previously been described as PASS, and should not be confused with aphids herein designated as R+ (= PAR-positive). U-type and T-type bacteria were not found in the single sample tested from California (Sandstrom et al., 2001).
At least Buchnera, PASS, and PAR, then, may co-occur in California pea aphids, and all are transovarially transmitted to offspring (Buchner, 1965; Chen & Purcell, 1997). Buchnera occur exclusively in specialised aphid cells, the bacteriocytes (= mycetocytes), and have not been detected in other aphid tissues. PASS may also be associated with bacteriocytes, in the sheath cells that surround them (Griffiths & Beck, 1973; McLean & Houk, 1973; van der Wilk et al., 1999; Fukatsu et al., 2000). Recently, in situ hybridisation with PASS- specific probes and electron microscopy have shown PASS, or R-type symbiont, to occur within bacteriocytes that until now have not been distinguished from the primary bacteriocytes containing Buchnera (Fukatsu et al., 2000; Sandstrom et al., 2001). PAR is easily detected in large numbers in the haemolymph and no intracellular association is as yet known, although all other known Rickettsia associated with insects are intracellular.
Many aphids are sensitive to high temperatures. For example, some species of Sitobion fail to reproduce at 28 °C in the laboratory (Turak et al., 1998) and pea aphids do not reproduce if subjected to a temperature of 37 °C for several hours as first-instar larvae (Ohtaka & Ishikawa, 1991). Rearing pea aphids for three generations at 25 °C also curtailed reproduction severely (Chen et al., 2000). Heat has been used with varying success to render various insects aposymbiotic (i.e. without symbionts) (Douglas, 1989), and it is reasonable to assume that in some cases, at least, high temperatures interfere with the reproduction of aphids via disruption of the primary symbionts (Ohtaka & Ishikawa, 1991).
Preliminary results (Chen et al., 2000) suggested that the presence of PASS or PAR might mitigate the deleterious effects of elevated rearing temperature on fecundity of the pea aphid. It was hypothesised that if the deleterious effects of elevated temperature on pea aphid reproduction were related to disruption of Buchnera, any ameliorative effect of facultative bacterial symbionts might be related, at least in part, to their interaction with the primary symbiont. Temperature effects on the primary symbiont Buchnera, on aphid fecundity, and on the incidence of PASS and PAR in natural populations, were investigated in a combination of laboratory and field studies.
Materials and methods
Aphids and plants
Two pea aphid clones that were free of PASS and PAR (P–R–) were used to establish, by injection of haemolymph from infected aphids, two sets of artificially derived subclones that were positive for PASS (P+R+), for PAR (P–R+), or for both PASS and PAR (P+R+), as described by Chen et al. (2000). Aphids were maintained at 18 ± 2 °C, LD 16:8 h, and were the same clones used by Chen et al. (2000). All aphids were reared on seedlings of sweet pea Lathyrus odoratus L. (cv. ‘Early multiflora mix’). On sweet pea, PASS had little effect on aphid fecundity, and PAR depressed fecundity by 11–16% (Chen et al., 2000).
Several pea aphid adults of a given clone (usually five adults per plant) were placed on pea seedlings and allowed to larviposit for 12 h. Larvae were kept at 18 °C for either another 24 h (referred to below as 1-day-old larvae) or for 5 days (5-day-old larvae), at which time they were counted and transferred to another growth chamber, where they were subjected to a period of heat stress. The temperature was monitored every 4–6 min using a Hobo® data logger and software (LogBook™ ver. 2.01, Onset Computer Corporation, Pocasset, Massachusetts). At the start of each experiment, the temperature rose gradually from 20 °C at a rate of 0.1–0.2 °C min−1 and reached 39 °C within ≈ 2 h. The entire experimental period was 6 h, so aphids were subjected to the maximum temperature of 39 °C for ≈ 4 h. The temperature then dropped gradually to 20 °C. The same growth chamber was used to generate heat-stress and cool-down periods for all experiments.
Heat-stressed aphids were kept at 18 °C and survivors were counted 2 or 3 days post-treatment. Aphids were monitored daily thereafter until they reached adulthood. Preliminary experiments indicated that > 90% of treated aphids that were capable of reproducing would do so by day 9 (C. B. Montllor, unpublished). Douglas and Dixon (1987) reported that maternal bacteriocyte numbers in pea aphids were highest in larvae, and remained relatively high over the first week of adulthood (70–80% of the maximum), then declined rapidly in older aphids. Therefore, in experiments for which fecundity and bacteriocyte number were both measured in the same individuals, adult aphid fecundity was monitored for 8 days, after which maternal bacteriocytes were counted.
Subsets of offspring of 1-day-old treated aphids and 5-day-old treated aphids were saved for subsequent testing for PASS. Of eight sets of tested offspring, all were P+ by polymerase chain reaction (Chen & Purcell, 1997), indicating that PASS survived heat stress and had been vertically transmitted to daughters successfully. The very few offspring of R+ parents were not tested for PAR.
In order to count bacteriocytes, adult aphids that had been heat-stressed as 1-day-old larvae were fixed overnight in FAA (70% ethanol 250, formalin 25, acetic acid 12.5) 8 days after the last moult. They were then rinsed and stored in 70% ethanol, and kept at 4 °C for subsequent dissection and bacteriocyte counts. Between 18 and 35 aphids of each clone (P–R–2 and its P+ and R+ subclones) were used. Between 12 and 15 aphids of each of the same three clones, maintained at a constant 18 °C, were also monitored for fecundity and fixed in FAA for bacteriocyte counts.
Fixed aphids were dissected in a drop of deionised water using fine needles to release bacteriocytes, and observed under a dissecting microscope. Bacteriocytes that were recognisable by their characteristic shape and colour were counted (Douglas & Dixon, 1987). Some treated aphids contained bacteriocytes that were much smaller than normal; if these were confirmed by dark-field microscopy (100–400×) to be bacteriocytes, they were also recorded.
Aphids were collected from alfalfa fields in at least four sites in each of three widely dispersed locations in California. Sites within locations were ≥ 1500 m apart. Overall, alfalfa-growing regions in three counties, representing a wide range of latitudes (range 32–40°N) and elevations (−10 to 1400 m), were selected as locations. Aphids were collected twice at each site within a location, in winter or spring (early) and in summer (late). All sites were in the vicinity of weather-collection stations that post daily temperatures on-line through the University of California Statewide Integrated Pest Management Project. At each site, ≈ 75 alfalfa stems were sampled using a canvas sweep net. Aphids were brought back to the laboratory where they were frozen and subsequently tested for PAR and/or PASS by PCR (Chen et al., 2000).
Fecundity of stressed and unstressed aphids of different clones was compared by computing 95% confidence intervals (t0.05SE) for each mean. Means with confidence intervals that did not overlap were considered significantly different (P < 0.05). Number of maternal bacteriocytes was analysed using anova and Newman–Keuls post hoc tests (Statistica® ver. 4.5, SI International, McLean, Virginia). The correlation between bacteriocyte number and fecundity (square of Pearson correlation coefficient) was analysed using the Bartlett χ2 statistic (Systat® ver. 5.2.1, SPSS Inc., Chicago, Illinois). Survival of different clones after heat stress and proportions of aphids with PASS or PAR on different collection dates were compared using the χ2 statistic (with Yates correction where appropriate) (Little & Hills, 1972).
Data from each of the two sets of parent and derived subclones were analysed separately and yielded similar results; significant differences (P < 0.05) are indicated by non-overlapping CI (Fig. 1). The fecundity of pea aphids was depressed significantly by heat stress for each set of clones. Bacterial status also affected aphid fecundity significantly within each set of clones. Aphids of both parent clones, without PASS or PAR, that had been heat-stressed as 1-day-old larvae essentially did not reproduce over the first 8 days of their adult life (one of 72 aphids produced a single larva); however 80–100% of the aphids containing PASS (P+R–) produced offspring, averaging 23.4 and 12.9 larvae respectively in the two different subclones. Although this reproductive output represented only 31–49% of the fecundity of aphids of the same clones reared at 18 °C, it was a significant increase over the output of heat-stressed P–R– aphids. Heat-stressed aphids with PAR, either alone or co-infected with PASS, did not produce significantly more offspring than aphids without either facultative symbiont (Fig. 1).
When 5-day-old aphids were heat-stressed, a similar pattern emerged with respect to the relative fecundity of clones of different bacterial status (Table 1). Fecundity was reduced most strikingly in P–R– aphids, and aphids with only PASS produced the most offspring, three to 14 times as many as their respective P–R– parent clones. For both sets of clones, the advantage of having PASS was decreased significantly when it co-occurred with PAR. Heat-stressed 5-day-old larvae also produced some apparently stillborn offspring as adults. Stillborn larvae were deposited with legs and antennae appressed tightly to the body, were immobile, and became desiccated while still attached to the mother or nearby. Stillborn larvae had never before been observed in the laboratory colonies. P–R– aphids, which produced the fewest live larvae, also produced more stillborn larvae than aphids with PASS or PAR, and this difference was significant for clone P–R–3 and its subclones (Table 1).
Table 1. Mean number of larvae (± 95% CI) born over 8 days to adults of parent clones P–R–2 and P–R–3 and their respectively derived subclones when heat-treated at 39 °C for ≈ 4 h as 5-day-old larvae (see text). (For P–R–3 and subclones, larvae of four groups of 10 adults each were counted.)
Number of live larvae
Number of stillborn larvae
Means followed by different letters within the same set of clones have non-overlapping CI and are significantly different (P<0.05).
For clone P–R–2 and its P+ and R+ subclones, maternal bacteriocytes were counted in 82 heat-stressed and 40 unstressed individuals for which fecundity data had been collected over 8 days (Fig. 2). In heat-stressed aphids of all clones, the mean number of maternal bacteriocytes in 8-day-old adults was reduced relative to unstressed aphids (F1,116 = 171, P < 0.01) but PASS-positive aphids retained significantly (up to nine times) more bacteriocytes than the other heat-stressed clones (Newman–Keuls test, P < 0.05) (Table 2). Bacteriocyte numbers were reduced by 93% in aphids that were P–R–, by 58% in P–R+ aphids, but by only 30% in P+R– aphids, compared with their unstressed counterparts (Table 2).
Table 2. Number of maternal bacteriocytes in 8-day-old adults of clones P–R–2 and two derived clones containing PASS (P+) or PAR (R+) (means ± SE). 18 °C: aphids reared at constant temperature; heat-stressed: 39 °C for ≈ 4 h as 1-day-old larvae (see text).
Number of bacteriocytes
Means followed by different letters are significantly different (anova, Newman–Keuls test, P<0.05).
Fecundity was not correlated significantly with numbers of bacteriocytes in unstressed P–R– aphids (R2 = 0.0005, χ2 = 0.001, 1 d.f., P = NS), P+R– aphids (R2 = 0.28, χ2 = 3.8, 1 d.f., P = NS), or P–R+ aphids (R2 = 0.29, χ2 = 3.2, 1 d.f., P = NS) (Fig. 2a); however, in heat-stressed aphids containing PASS, there was a significant, positive correlation between numbers of bacteriocytes and numbers of larvae born over the first 8 days of adulthood (R2 = 0.31, χ2 = 9.8, 1 d.f., P < 0.01) (Fig. 2b). Only three of 18 PAR-only aphids and one of 35 P–R– aphids produced any larvae, therefore a correlation was not attempted for these clones.
Survival of 1-day-old larvae, with or without facultative symbionts, subjected to heat stress was determined within 3 days after treatment for two experiments. For analysis, aphids of the same bacterial status from both sets of clones (P–R–2 and P–R–3 and their respective subclones) were pooled. Aphids with PAR only had the lowest survival (23/81 = 28%) while aphids with both PASS and PAR had significantly higher survival (93/180 = 52%). Survival was 78/160 (49%) for P–R– aphids and 57/138 (41%) for P+R– aphids (χ2 = 13.9, 3 d.f., P < 0.01). Though survival per se was not measured subsequently, there were fewer PAR-only aphids further along in the experiment for both treated 1-day-old and 5-day-old larvae, such that R+ clones were under-represented in experiments.
Natural populations of aphids had a significantly higher proportion of PASS on late compared with early collection dates at two of the three locations (Table 3). As an estimate of differences in ambient temperatures influencing early and late pea aphid populations in the field, the mean daily high temperature was calculated for the month preceding each collection. The number of days below 20 °C preceding the early collection, and above 30 °C in early or late collections, was also calculated (Table 3). (According to one study, the optimal temperature for pea aphid reproduction is 10–20 °C, and the upper threshold for pea aphid development is ≈ 28 °C; Campbell & Mackauer, 1977.) The proportion of aphids with PASS was significantly higher at the San Joaquin and Imperial Co. sites (χ2 = 25.0, 1 d.f., P < 0.01; χ2 = 4.0, 1 d.f., P < 0.05 respectively) in late compared with early collections. At the Lassen Co. sites, there was a more than five-fold increase in frequency of PASS in late compared with early collection but the difference was not significant (χ2 = 2.9, 1 d.f., P = NS, with Yates correction) (Table 3). The incidence of PAR was low on all dates (0–9%) and no trend was evident relating the frequency of PAR and date of collection (data not shown).
Table 3. Temperature parameters for month preceding each collection, and frequency of PASS in samples of natural populations of pea aphids, from three locations in California. Early = winter/spring 2000, Late = summer 2000, n= number of aphids assayed.
Mean daily high (range) (°C)
Days <20 °C or >30 °C
Frequency (%) of PASS (n)
Mean daily high (range) (°C)
Days >30 °C
Frequency (%) of PASS (n)
Frequency is significantly different between collection dates (χ2, P<0.05).
The presence of a facultative bacterial associate of the pea aphid, PASS, enhanced the reproduction, though not necessarily the survival, of aphids subjected to heat stress as very young (1-day-old) or as older (5-day-old) larvae. This effect could be related to PASS rescuing the primary symbiont, Buchnera, which was also more sensitive to heat stress in the absence of PASS. Reproductive output and numbers of bacteriocytes were correlated positively in heat-stressed aphids, but the possibility that PASS affected reproduction of aphids and survival of Buchnera independently of each other cannot be ruled out.
The increase in PASS-positive pea aphids in the natural populations sampled in warmer compared with cooler weather at the same sites is consistent with the hypothesis that PASS benefits its aphid hosts in the field. Temperature, however, was probably not the only factor involved in determining the frequency of PASS. The highest incidence of PASS did not occur in the hottest region, Imperial Co., with the highest average and maximum temperatures. There was no correlation between average temperatures and incidence of PASS among four locations in California, including the three discussed in this study and a fourth site in Humboldt Co., sampled on a total of seven dates from February to August 2000 (data not shown).
Although this study showed a clear advantage to pea aphids that harbour PASS under environmental conditions that occur in summer in parts of California, work to date suggests that PAR is pathogenic to pea aphids (Chen et al., 2000), and its existence in pea aphid populations is not yet explained by any putative benefits. Pea aphid rickettsia (PAR) did not ameliorate the effects of heat-stress on aphid fecundity or bacteriocyte numbers, and the incidence of PAR did not increase significantly in late (summer) collections of pea aphids in the field (data not shown). Furthermore, the presence of PAR often negated the beneficial effects of PASS to heat-stressed aphids in this study. Whether pea aphids may derive benefits from PAR under a different set of environmental conditions than they do from PASS, as suggested more generally for multiple infections by Douglas (1998b), has yet to be shown experimentally.
The numbers of bacteriocytes counted in pea aphids without facultative bacteria corresponded to previously reported data from 8-day-old pea aphid adults (≈ 60 maternal bacteriocytes; Douglas & Dixon, 1987). Clonal variation in numbers of maternal bacteriocytes in pea aphid adults, and a positive correlation between bacteriocyte numbers and aphid weight have been reported (Wilkinson & Douglas, 1998). The lower number of maternal bacteriocytes in an untreated pea aphid clone infected artificially with PAR (the R+ subclone of P–R–2) is suggestive of a pathogenic effect of PAR, and should be followed up with an examination of naturally infected pea aphid clones.
The importance of density or titre of symbiotic bacteria for the performance of aphids has not been studied extensively. Humphreys and Douglas (1997) investigated the effect of rearing temperature on density of Buchnera in pea aphids. They reported that aphids reared at 25 °C were smaller that those reared at 15 °C, and had a higher density of Buchnera in embryonic but not maternal tissues, but fitness at these different temperatures was not measured. If > 70% of Buchnera occur in embryonic compared with maternal tissues, as reported (Whitehead & Douglas, 1993; Humphreys & Douglas, 1997), the estimated differences in Buchnera density assumed in the present study are quite conservative, because treated aphids with fewer maternal bacteriocytes also had fewer embryos (data not shown) and hence fewer embryonic bacteriocytes.
The effects of the intracellular symbiotic bacterium Wolbachia were correlated with bacterial density, whether it varied naturally or was manipulated experimentally by antibiotic treatment (Boyle et al., 1993; Breeuwer & Werren, 1993; Zchori-Fein et al., 2000). In this study, the density of Buchnera was manipulated essentially by exposing host aphids to different temperatures. It is clear that an experimental manipulation that reduced the number of bacteriocytes (and hence, presumably, the Buchnera population), i.e. a heat-stress treatment, lowered the fecundity of pea aphids severely. Conversely, the aphid clones that were able to retain more bacteriocytes (PASS-positive clones) after heat stress were affected less severely than clones with fewer bacteriocytes. In addition, fecundity was correlated positively with numbers of bacteriocytes within the heat-stressed clone that was PASS-positive. A question that remains, however, is how the presence of PASS is related causally to bacteriocyte numbers. It is also not known how populations of the facultative bacteria were affected by heat stress, although it was clear that PASS survived the treatment. Bacterial densities may also explain partially why experimental results with PAR were more variable in both this and a previous study (Chen et al., 2000). PAR populations may be under less stringent host control than those of PASS, resulting in more variable levels of this particular symbiont.
Counting maternal bacteriocytes is only a rough measure of Buchnera populations. Since the discovery of secondary mycetocytes (= bacteriocytes) that contain mostly or exclusively PASS, and not Buchnera (Fukatsu et al., 2000), it is not clear to what degree these might account for the greater number of bacteriocytes counted in PASS-positive heat-stressed aphids. Sandstrom et al. (2001) also reported that two types of secondary symbionts, including one presumed to be virtually the same as PASS, were found in cells resembling bacteriocytes that did not contain Buchnera. These authors examined only aphid embryos. Fukatsu et al. (2000) reported that secondary bacteriocytes were not normally part of the maternal complement of bacteriocytes; rather, they were typically found in embryos, and comprised a small percentage of the total number of bacteriocytes, however it is also possible that PASS might replace Buchnera in bacteriocytes of heat-stressed aphids. This issue must be resolved by microscopy of heat-stressed aphids.
It is becoming increasingly clear that biological studies of insects are complicated by the interactions of several genomes (Heddi et al., 1999). For pea aphids, at least nine transovarially transmitted genomes are potentially involved in the physiology of the host: nuclear, mitochondrial, Buchnera, PAR, PASS, U- and T-type symbionts (Sandstrom et al., 2001), a Spiroplasma symbiont (Fukatsu et al., 2001), and a phage of secondary symbionts (van der Wilk et al., 1999; Sandstrom et al., 2001). To date, U- or T-type bacteria, Spiroplasma, and phage have not been reported from California pea aphids. Ultimately, the presence or absence of all known facultative symbionts and the interactions among the various genomes must be taken into account in order to understand fully the ecology of natural populations of pea aphids.
We are indebted to L. Goldstein, C. Silva and K. Tansavatdi for technical assistance, to F. Barre and P. Barre for statistical advice, and to F. Barre and T. E. Mittler for critical reading of the manuscript. This work was supported in part by USDA competitive grants 96-35302-3544 and 00-35302-9400.