We conducted field surveys and experiments to evaluate the hypothesis that predation is an important driving factor determining the degree of coexistence between red and green morphs of the pea aphid Acyrthosiphon pisum. Theory suggests that the different colour morphs are differentially susceptible to natural enemies and selection by predation which in turn leads to variable relative abundances of red and green morphs among host plants across landscapes. Our field surveys on pea and alfalfa revealed, however, that the colour morphs tended to coexist closely in a ratio of one red to three green aphids across fields with different host plant monocultures. Experimentation involving manipulation of the relative abundances of the two colour morphs on host plants pea and alfalfa with and without predator presence revealed that red morphs had higher or same fitness (per capita reproduction) than green morphs on both pea and alfalfa only when in the proportion of one red/three green proportion. Moreover, experimentation evaluating predator efficiency revealed that red morphs are safest from predation when in a 1 : 3 ratio with green morphs. These results suggest that in addition to predation selection effects, red morphs may behaviourally choose to associate with green morphs in a narrow 1 : 3 ratio to maximize their fitness. This evidence, along with existing published data on red and green morph anti-predator behaviour indicates that a 1 : 3 red and green morph coexistence ratio is driven by a balance between predation pressure and behavioural assorting by red morphs across landscapes. In this way predators may have ecological-evolutionary consequences for traits that affect the colour morphs' proportion and tolerances to selective pressure.
Phytophagous insects – and especially the pea aphid complex (Acyrthosiphon pisum Harris) – are useful models to identify how divergence in plant host use and aphid colour polymorphisms may have arisen (Simon et al., 2003; Sword et al., 2005; Kunert et al., 2010). Based on study of 1101 base pair region of the mitochondrial cytochrome oxidase I gene (COI) sequenced for 21 pea aphid clones (collected from Canada and USA) Boulding (1998) rejects the idea that there are host races of the pea aphid. Analyses based on genetic (DNA) markers however confirm the existence of 11 sympatric pea aphid populations in Europe all adapted to different host plants from the family Fabaceae (Peccoud et al., 2009a). Most notable are populations adapted to alfalfa – Medicago sativa – and to red clover – Trifolium pretense – which are now considered sympatric races because of the high level of genetic divergence between them. Both can feed and reproduce normally on their own host but cannot normally feed and reproduce on other host plants in the family Fabaceae (Simon et al., 2003; Peccoud et al., 2009a,b).
Within each host adopted pea aphids, there are red and green colour morphs that coexist on their respective host plants. The basis of the colour polymorphism between red and green morphs appears to be essentially genetic and is based in a single carotenoid desaturase encoded only by red morphs (Moran & Jarvik, 2010). The colour is stable throughout parthenogenetic reproduction such that green females only give birth to green offspring, red females to red offspring. Thus, the two colour morphs coexist during their asexual reproduction as though they were two sympatric races (Via, 1991, 1999; Via et al., 2000; ; Simon et al., 2003; Peccoud et al., 2009a,b; Moran & Jarvik, 2010).
The divergence in host use by the colour morphs is believed to have originated on a common broad bean (Vicia fabae L.) host (Via, 1991, 1999; Via et al., 2000). One potential driving factor for host divergence in pea aphid colour morphs is susceptibility to natural enemies (Losey et al., 1997).
Many predators, including ladybird beetles (Coccinelids), damsel bugs (Nabids), hover flies (Syrphids) and hymenopterous parasitoid wasps (Braconids) are known to attack pea aphids (Snyder & Ives, 2003; Nelson & Rosenheim, 2006; Nelson, 2007). More specifically, Losey et al. (1997) showed that the green-red colour polymorphism may be maintained by differential vulnerability to natural enemies because green morphs are most susceptible to parasitoids and red morphs are most susceptible to predators (Losey et al., 1997). Moreover, red and green morphs should vary in relative abundance depending on natural enemy presence and relative density of red and green morphs. This study was conducted on a single host plant alfalfa and so didn't reveal the potential costs and benefits of predator effect on genetically different host races. Some studies however mention different proportions (i.e. 100% of green in pea) of red and green morphs of pea aphids on pea, cover and alfalfa at the beginning of the host plant grooving season (Simon et al., 2003; Frantz et al., 2006, 2009), neither of these followed specifically the survey of colour morph and neither aforementioned covers the whole vegetative growing period. Moreover, the relative performance of green and red morphs on different host plants under predator pressure remains completely unknown. In particular, the colour morphs also exhibit different propensities to engage in escape behaviour (red morphs drop off the host plants more frequently than green morphs when predators are present (Braendle & Weisser, 2001). This may be because red morphs are more susceptible to predation than green morphs, such that there is a marked response of red morphs to ladybird kairomones associated with crowding (Dixon & Agarwala, 1999). These studies show that the colour morphs are under different predator/parasitoid pressure and that this is also reflected in their respective behaviours. This avoidance behaviour may however have fitness consequences: red individuals that spend more effort avoiding predators feed less and thus could suffer reductions in offspring production. Even so, the red morph's susceptibility to predation may vary with the relative abundance of green and red colour morphs that coexist on a host plant. This implies that coexistence on a plant may be driven by the red morphs' propensity to associate with green morphs to reduce predation risk, rather than merely through differential mortality via predation. If this is so, one would expect some convergence to fixed narrow ratios of red and green colour morphs coexisting on host plants. On the basis of these previous observations, we hypothesize that some proportion of the two colour morphs on different host plants may have exits and predation may be an important driving factor determining the degree of coexistence between red and green morphs of the pea aphid. If so this may have fitness benefits and may be evolutionary stable.
We explored this possible mechanism for coexistence using a series of field surveys and manipulative experiments. An intensive field survey was used to establish the ratios of colour morphs on host plants under field conditions in agricultural ecosystems in Europe. Subsequent field cage experimentation was performed to determine if the observed ratios represented conditions that lead to a higher relative fitness of red compared with green colour morphs. Predator direct preferences of colour morphs in different combinations were also tested.
Materials and methods
All work here described was conducted according to relevant national and international guidelines. Before aphid collection, farmers were contacted and permission was obtained to enter their fields. For insect collection, no permits were required as the area where pea aphids and lady beetles were collected did not contain any strictly protected areas. Permits were also not required to use insects for experiments due to the observational nature of the data collection.
Field assessment of red/green ratios
Two alfalfa and two pea (Pisum sativum L.) fields of about three ha each were surveyed for field experiment. The alfalfa fields were separated by approximately 350 km and the pea fields were separated by 210 km. This ensured that we were most probably dealing with genetically distinct populations. Weekly assessments of colour morph relative abundance were made in each field from May until the end of September (harvest). On each sampling date six plots of about 6 square metres each were randomly established within each field. Within each plot, another six subplots of about 1 square metre each were established (Fig. 1a). Inside each subplot ten plants were randomly selected and brushed into a plastic dish. Pea aphids were counted by evaluating green and red morphs separately. The total numbers of red and green morphs from ten plants were considered and noted separately. Aphids were then released outside of the plot and the procedure was repeated for the next subplots. The entire procedure was repeated 1 week later by randomly establishing another six plots and subplots inside the same fields. Altogether 36 samples/field/week were collected. A massive yield loss due to heavy leaf miner fly infection in mid-summer precluded data collection in the other pea field.
During the survey all predators were counted by separately evaluating the most frequent species. Two-spotted ladybird beetles – Adalia bipunctata L.– were the most frequent predator during the whole assessment period, their number reaching a maximum of 44 individuals in ten plants of one subplot. Thus, ladybirds were used in the manipulative experiments. Individuals for experimentation were obtained from the same fields 48 h prior to use in cages.
Fitness consequence assessment of red/green ratios
The field experiment was conducted using pea and alfalfa host plants in a typical agriculture planting configuration i.e. crop rows within one of the previously assessed alfalfa and pea fields. The experiment comprised 100, 80 cm wide × 120 cm high enclosure cages per field. Each cage was placed over 10–12 plants (Fig. 1b). All insects including aphids were removed systematically from host plants inside cages every day for 1 week with an aspirator. Alfalfa and pea plants near to the cage margins were also removed up to approximately 15–20 cm from the margins. This was done to prevent aphids and predators from climbing up the cages. To initiate the experiment, four adult pea aphids – previously collected from the same field – were added in the following combinations:
– 1 red + 3 green (R1G3) (ten predation, ten control cage)
– 2 red + 2 green (R2G2) (ten predation, ten control cage)
– 3 red + 1 green (R3G1) (ten predation, ten control cage)
– 4 red + 0 green (R4G0) (ten predation, ten control cage)
– 0 red + 4 green (R0G4) (ten predation, ten control cage)
These ratios bracketed those observed under field conditions (see Results). Aphids were left to establish on host plants, feed and reproduce for 4 days, having approximately six to nine offspring per female per day. At this time, predators were stocked to half of the cages and the remaining half were used as predator-free controls (Fig. 1b).
Adalia bipunctata, the dominant species of ladybird beetle predator in the field surveys, were collected from the fields and placed into plastic boxes separate from alfalfa and pea fields. They were obtained 48 h prior to use for experiment. If mating between beetles was observed they were isolated and later used together in cage experiments (1 female and 1 male for per cage). Stocking of males and females in the same cages was done since the predation level of male and female ladybirds may differ. Mated females in particular are considered to have voracious appetites during egg formation. Because some plants were attended by ants from the soil, these were excluded completely from any further assessment. The whole experiment lasted from 10 to 12 days. At that time female beetles laid eggs and additional new emerging lady beetle larvae would lead to highly differential predation rates among treatments. Thus, when eggs were observed in predator treatments (first time after 10 days) the enclosure cages were carefully removed and the whole host plant cut down, and the insects on it brushed into a white plastic dish. All surviving aphids were counted, and green and red morphs assessed separately.
Predator direct preferences assessment of red/green ratios
Adult pea aphids and their ladybird (A. bipunctata) predators were collected from pea and alfalfa fields prior the experiment. Male and female predators were held in cages the same way as described for the cage experiment. Four aphids were placed in 9 cm diameter covered Petri dishes in the same colour combinations used for the cage experiment. The background colour of the Petri dishes was green to simulate the natural environment. A randomly chosen (male and female) ladybird adult was placed in a Petri dish and left to move around and feed. If the predator made a clear choice and started to eat one of the aphids the experiment was stopped, the colour option of predator and time elapsed until the predation event was noted. The maximum elapsed time for a feeding trial was 15 minutes the time standardized for all feeding trials. The experiment was replicated 20 times for each aphid colour combination. The entire procedure was repeated for aphids collected from pea fields and alfalfa fields. The entire experiment was conducted at the end of August during normal summer time conditions from 11:00 to 16:00 hours.
The colour ratios of pea aphid red and green morphs for all sub plots (samples from ten plants) were separately analysed and the frequencies of colour ratios/ten plants for all fields were considered separately.
The fitness consequences of red/green ratios were analysed by comparing the performance of red and green morphs of all combinations in the presence and absence of predators. This was done by counting the per capita reproduction rata of adults reporting the number of survived offspring at the end of the experiment to the initial number of adults for each colour morphs. The first comparison evaluated the performance of red and green morphs in combination of 1 red 3 green starting number of adults on pea and then in alfalfa in the presence and absence of predator. Similar methods were used for the other colour combinations (R2G2 and R3G1) on pea and alfalfa. In case of combinations R0G4 and R4G0 predation treatment from each colour morph were compared with control and then a cross comparison between red and green in pea and then on alfalfa were also computed. For predator direct preferences the average time until a clear predation on red and green morphs was considered for all colour combinations.
ANOVA was used to test for fitness (cage experiment) and predation rata (preference experiment) of colour morphs on host plants. ANOVAs that revealed significant effects were followed by Tukey tests to identify significantly different treatments. Confidence limits of P ≤ 0.05 were considered significant.
The relative proportion of red and green morphs varied among plots, but the highest frequency colour ratio was 25–30% red/70–75% green (or approximately 1 red/3 green) on both pea and alfalfa fields (Fig. 2). Green morphs dominated the pea host plants (100% green, 0% red morphs) only from the beginning of the growing season until the end of June. The highest proportion of red morphs was 45–50%; however, this occurred at low frequency only (Fig. 2). The field assessment never revealed the existence of 3 red/1 green ratio (70–75% red and 25–30% green) nor 4 red/0 green (100% red and 0% green) ratios.
The field experiment involving pea plants showed that red morphs under predator pressure had the same fitness as greens both under predation (F = 2.06, P = 0.16) and under no predator pressure (F = 2.42, P = 0.06) when the morphs were in a ratio of 1 red 3 green (Fig. 3a). Red morphs under predation had lower fitness than green morphs when coexisting in either a 2 :2 red: green ratio (F = 3.15, P = 0.03) or a 3 : 1 red:green ratio (F = 5.80, P < 0.001) (Fig. 3b, C). ANOVA revealed no significant differences in fitness when both colour morphs existed alone (predation F = 0.5, P = 0.96; control F = 1.47, P = 0.30) (Fig. 3d).
Neither of the colour combinations had fitness loss on green morphs on host plants pea (Fig. 3a, b, c, d).
ANOVA revealed that red morphs on alfalfa under predator pressure had higher fitness than green morphs when in a 1 : 3 red:green ratio (F = 6.25, P < 0.001) but no fitness differences occurred in the control (F = 0.34, P = 0.80) (Fig. 4a). Lower fitness was observed when in a 2:2 ratio under predation (F = 2.71, P = 0.05), and no difference in control (F = 1.76, P = 0.22). Also red morphs had lower fitness than greens when in a 3 : 1 red:green ratio under predation (F = 13.23, P < 0.001) (Fig. 4b, c). The comparison of pea aphid colour morphs performing alone revealed lower fitness of red aphids when performed alone on alfalfa (F = 3.40, P = 0.02) (Fig. 4d). Fitness loss of green morphs on alfalfa was observed only under predation in 1 red : 3 green proportions (Fig. 4a) and a clear fitness benefits was revealed in any other colour combinations under predation (Fig. 4b, c, d).
Predator attack on red and green morphs was significantly different between 1 red : 3 green and 3 red : 1 green ratios. On both pea and alfalfa it took approximately three times longer for predators to capture the reds in 1 red : 3 green proportion whilst the predator attack on greens was two to three times higher only in 3green : 1 red proportions (Fig. 5a, b).
Other studies shows that green, red, white, orange, yellow and mist colour polymorphism exists in Macrosiphoniella yomogicola aphid population and that this high variety may be maintained because selection may favour high colour diversity (Agawa & Kawata, 1995).
The functional origins of pea aphid red and green colour polymorphism and their coexistence on host plants have been explored from both a genetic and population dynamic perspective (Losey et al., 1997; Simon et al., 2003; Peccoud et al., 2009a,b; Moran & Jarvik, 2010). However, neither approach has explored the potential fitness consequences (both costs and benefits) of individuals when coexisting in different ratios on host plants.
Our whole vegetation study revealed that the most frequent ratios of red and green morphs on both pea and alfalfa are very close to 1 red : 3 green (Fig. 2). Moreover, like Simon et al. (2003), we only observed green colour morphs existing in isolation (i.e. 100% of green on pea) at the beginning of the vegetation season on May until June. Other studies made in Serbia also demonstrated that the first collected pea aphid colour morphs at the beginning of the vegetation period were greens (Tomanović et al., 1996). The same study also reported that red morphs reached the maximum rate of 47.3%. This result is very appropriate with ours and further demonstrates that red-green fixed proportion is an evolutionary stable strategy for pea aphids. Because geographical separation among populations within our study was large, we can rule out local adaptation driving the tight frequency in colour morph ratios.
Assessing the total colour ratios at the end of the field assessment by counting all individuals collected throughout the season we observed a very similar 1 to 3 proportion of red and green morphs for all fields (from 2099 individuals on first alfalfa field 23.61% red and 76.39% green; from 3419 individuals on a second alfalfa field 25.78% red and 74.22% green and from 5876 individuals on pea 24.12% red and 75.88% green).
Our experiment, which manipulated the colour morph ratios on host plants and examined fitness effects in the presence and absence of predation, provides some clues as to a possible mechanism for the narrow coexistence conditions in the field. In the face of ladybird predation, red morphs had similar fitness as green morphs on pea plants only when in a 1 red : 3 green ratios and no fitness loss on green morphs were revealed (Fig. 3a, b, c, d). Red morphs on alfalfa had higher fitness than greens in when in the same ratio (Fig. 4a, b, c, d). In all other cases, green morphs had higher fitness. One important factor in colour morphs selections is visual predators (Bond, 2007). According to Bond (2007) for predators it may be harder to search simultaneously for two cryptic prey types than to search for one. Therefore, visual predators should tend to focus on the less cryptic colour morphs and neglect the others. Predators in this way may induce frequency-dependent selection and stabilize the prey colour polymorphism. Our results offer one of the first evidence of this hypothesis. Higher predation risk on red morphs is also demonstrated in our predator preference trials, but only when red morphs and green morphs coexist in ratios other than 1 : 3. This suggests that predators may have difficulty distinguishing between colour morphs as potential prey when the relative abundance of the red colour is low. It is possible that the outcome of the predator preference experiment may not be a result of the preference of the predator but due to the different abundance of the two colour morphs. If there were 4 aphids in a Petri dish, 3 red and 1 green, it might be faster for the ladybird to get a red aphid simply because they are more abundant. This however can also suggest that for red morphs the higher proportion than 1 : 3 is not suitable and can explain why we never found 3 red : 1 green or only100% red on field.
Different selective pressure however indicates also different behavioural aspects. Colour morphs exhibit different escape behaviour. Red morphs drop more frequently than green morphs when predators including ladybird beetles are present (Braendle & Weisser, 2001). Red morphs also differed significantly in a late production; red clones produced on average a higher proportion of winged morphs than green clones (Weisser & Braendle, 2001). Winged morphs are produced under stress (predator pressure) (Balog et al., 2013 in press) reducing in this way the predator pressure by new – if possible enemy free – host plant colonization (Weisser & Braendle, 2001; Kunert et al., 2010). This further indicates that different selective pressure exists between red and green colour morphs. Chemical cues may also be important selective force, and one other factor that may influence pea aphids is the parasitism (Tomanović et al., 1996; Losey et al., 1997). Green morphs of pea aphids are generally more parasitized than red morphs (Tomanović et al., 1996; Losey et al., 1997). This difference in parasitism may be because green morphs are more frequent than the red morphs and also because predators and parasitoids avoiding this way intraguild predation (Losey et al., 1997).
Our study suggests that red morphs assort themselves with green morphs among host plants across the landscape in coexistence ratios that minimize fitness loss to predation. Because the red colour on green foliage is noncryptic, the highest disturbance on red morphs in any other than 1 : 3 ratio may have direct consequences for red morph feeding process. A longer time spent engaged in predator avoidance – moving or searching new plants – could significantly reduce the fitness (per capita reproduction) of the reds morphs. This loss of fitness can be overcome by associating with green morphs in ratios that lower predation risk and thus maintain a fitness level comparable with green morphs. This behavioural assorting into a 1 red : 3 green ratio may reflect an evolutionary stable strategy that maintains the colour polymorphism in nature.
This work was supported by a grant of the Romanian National Authority for Scientific Research, CNCS – UEFISCDI, project number PN-II-RU-TE-2011-3-0096. Authors have no conflict of interest to declare.
AB and OJS conceived and designed the experiments. AB performed the experiments. AB and OJS analysed the data. AB and OJS wrote the manuscript.