Spatial and temporal cooccurrence among Neotropical native coccinellids and the exotic Harmonia axyridis

B.V


Introduction
Coccinellids are important aphid predators in horticultural crops around the world and are used in classical (Rondoni et al., 2021) and augmentative (Albajes et al., 2002) biological control programs. In the Neotropical region several species have deserved attention as potential biological control agents mainly in horticultural crops (Dode & Romero Sueldo, 2013;Dos Santos et al., 2020;Fonseca et al., 2017;Zalazar & Salvo, 2007).

3
Vol:. (1234567890) connexa and C. quadrifasciata octodecimpustulata have Neotropical distribution (González, 2014), and these species are present in several crops such as corn, sunflower, alfalfa, fruit trees, sweet pepper, coriander (Andorno et al., 2015;De la Fuente et al., 2006), recently in eggplant (pers. obs.), and weeds (Castresana & Paz, 2019;Engel et al., 2020;Saini, 2004). The first two coccinellid species can be considered good aphid control agents as they have high consumption rates on aphid species frequently found in horticultural crops (Rocca et al., 2017(Rocca et al., , 2021. C. quadrifasciata octodecimpustulata is a more generalist predator, feeding on immature and adult prey, including aphids, thrips, scale insects, psyllids, whiteflies, mites, and eggs and immature stages of coleopterans and lepidopterans (Lixa et al., 2010), and its potential as biological control agents have not been studied. On the other hand, H. axyridis, native to Asia, is widely distributed worldwide, and it has been introduced as a biological control agent or accidentally in many countries (Camacho-Cervantes et al., 2017;Roy et al., 2016). Due to several traits-larger size than other coccinellids, high predatory capacity, high fecundity and fertility, superior competitive ability in relation to other coccinellid species by exploitation and interference-H. axyridis became an invasive exotic species and has the potential to displace several native coccinellids (Brown & Roy, 2018;Roy et al., 2016). In South America, there are reports of such displacements of native species in Chile and Brazil (Grez et al., 2016;Martins et al., 2009). In Argentina this species was introduced in 1986 ( García et al., 1999) and nowadays it extends from 24° to 43° S and from 71° 30´ to 54° W, having arrived in ten biogeographical regions (Wagner et al., 2017;Werenkraut et al., 2020). Although there are no recent records of relative abundances of H. axyridis in relation to other coccinellids in horticultural crops, Saini (2004) found, in pecan trees, relative percentages between 51 and 74% of H. axyridis in relation to Cycloneda sanguinea, Olla v-nigrum, Eriopis connexa, Coleomegilla quadrifasciata y Adalia bipunctata.
The temporal and spatial cooccurrence of different species of coccinellids creates opportunities for intraguild predation that could negatively affect their coexistence (Burgio et al., 2002;Michaud, 2002;Schellhorn & Andow, 1999). This interaction is usually bidirectional and asymmetric (Michaud & Grant, 2003;Ware & Majerus, 2007), and its outcome is determined by several factors such as specific traits associated with particular species-e.g., physical and chemical defenses-, developmental stages, relative body size, mobility, behavior, and encounter rates of the individuals who participate in the interaction (Lucas, 2012;Lucas et al., 1998). In turn, eggs, small larvae, pupae, and molting stages are especially susceptible to intraguild predation (Hodek & Evans, 2012). It is known, for example, that H. axyridis exerts a strong asymmetric intraguild predation on eggs and larvae of C. sanguinea (Michaud, 2002) and E. connexa (Mirande et al., 2015), and Rocca et al. (2017Rocca et al. ( , 2019 found asymmetric intraguild predation between E. connexa and C. sanguinea, the former being the strongest intraguild predator on eggs and larvae of C. sanguinea. However, the interaction between C. quadrifasciata octodecimpustulata and the other three species is unknown. The knowledge of the cooccurrence of C. sanguinea, E. connexa, C. quadrifasciata octodecimpustulata, and H. axyridis in horticultural crops, as well as the intraguild predation between them, is relevant to develop efficient biological control programs involving native biological control agents in the presence of exotic species. We studied the spatial and temporal cooccurrence of these coccinellid species in eggplant crop, Solanum melongena L. (Solanaceae), which is economically important in many countries of the world and in Argentina (CHBA, 2005). This crop, like many other horticultural crops, is affected by the aphids Aphis gossypii Glover and Myzus persicae (Sulzer) (Del Pino et al., 2012;Rizzo, 2020), that extract the sap from the plants and reduce their photosynthetic capacity, deform the tender shoots, produce honeydew leading to the development of sooty mold, and most important, they are vectors of viruses (Srinivasan, 2009). The aims of this work were: 1) to describe the abundance of H. axyridis, C. sanguinea, E. connexa, and C. quadrifasciata octodecimpustulata, and the spatial and temporal cooccurrence of these species in the eggplant crop, and 2) to estimate, in the laboratory, the intensity and symmetry of the IGP between C. quadrifasciata octodecimpustulata and the other coccinellid species. Based on the aforementioned background, we predict that H. axyridis will have greater abundances in relation to the other coccinellid species, and in interaction with C. quadrifasciata octodecimpustulata it will be the strongest intraguild predator.

Field sampling
Biweekly samplings were carried out in four eggplant crops (250 m 2 each) located in agroecological farms of the Horticultural Belt of La Plata (34º 8' S, 57º 54' W), Buenos Aires, Argentina, during two production cycles from November to May (2015May ( -2016May ( and 2016May ( -2017. On each sampling date, the phenological stage of the crop was recorded: vegetative (V), flowering (F), flowering and fruiting (FF), fruiting (FR) and postharvest (P). The sample unit consisted of a plant that was visually divided into upper, middle, and lower stratum, all of them of equal size. On each sampling date, a visual inspection of 30 plants randomly selected was carried out, and the frequency of each coccinellid species (larvae and adults) by plant stratum was recorded. The relative abundance of the coccinellid species in each phenological stage (number of individuals of one species / total number of individuals in a certain phenological stage × 100), and each plant stratum (number of individuals of one species / total number of individuals in a given stratum × 100), was calculated. In turn, on each sampling date, 30 leaves per stratum were taken at random from different plants (n = 90 from each plant) and the number of aphids was recorded. The mean number of aphids in each phenological stage was estimated taking into account the sample units of all sampling dates corresponding to the same phenological stage and all sites. The mean number of aphids per stratum was estimated in the same way. Kruskal-Wallis test was used to compare, separately, the mean number of aphids among vegetative stages and strata.
The cooccurrence of the coccinellid species were determined at spatial (plant) and temporal (phenological stage) level using the co-occur package (Griffith et al., 2016) with R software (version 3.5. 1). The probabilistic model uses combinatorics to determine the probability that the observed frequency of cooccurrence is significantly equal to that expected (random association α = 0.05), greater than expected: P (gt) ≤ 0.05 (positive association), or significantly lower than expected: P (lt) ≤ 0.05 (negative association) of a pair of species (Veech, 2014).
For the cooccurrence analysis at the plant scale, all sample units were considered, from all sampling dates and sites, in which at least one of the coccinellid species was present. Regarding the temporal analysis, all sample units were considered, from all dates and sites, of the same phenological stage in which at least one of the species was present.

Laboratory assays
The insect rearing and all laboratory experiments were carried out under controlled environmental conditions (25 ± 2ºC, 70 ± 5% HR and 16:8 L:D). Coccinellids (H. axyridis, C. sanguinea, E. connexa, and C. quadrifasciata octodecimpustulata) were collected from agroecological eggplant crops in La Plata, Argentina (34°56′04″S 58°10′14″W). The progeny of each species was used to start the laboratory colonies. The adults and larvae were reared in separate plastic cages (15 cm high × 15 cm long × 25 cm wide) containing water provided on a sponge inserted into an Eppendorf tube, Triticum aestivum L. seedlings infested with Rhopalosiphum padi (L.) (Hemiptera: Aphididae), a paper towel to provide refuge for larvae and oviposition substrate for females and covered with a fine mesh material. Seedlings were previously germinated in plastic pots (6 cm high and 4 cm diameter) with standard substrate (fertile soil and perlite 1:1 v/v) and infested with R. padi at germination and maintained in ventilated plastic boxes (13 cm high × 30 cm long × 23 cm wide) until more than 80% of each were infested with aphids. Twice a week, new seedlings with aphids were added and the paper towel with eggs from the adult cages was transferred to a new plastic box until the larvae hatched. Individuals were randomly selected from the colonies and starved for 24 h prior to the experiments.
The experimental unit consisted of plastic cylinders (6 cm diameter × 5 cm high), containing an eggplant leaf embedded in water agar (1%), sealed with voile. On the leaf, 5 M. persicae adults were placed as extra-guild prey, simulating a situation of scarcity of prey.

IGP by adults and larvae on eggs
One adult reproductive female coccinellid of 2-3 weeks old of each species was individually isolated in the experimental units with a heterospecific egg cluster (20-30 eggs each). Similarly, one second instar larva (L2) or one fourth instar larva (L4) of each coccinellid species were confined with a heterospecific egg cluster. After 24 h, the number of eggs consumed was recorded. The number of predated eggs was analyzed as a binary response variable (consumption/non-consumption) using a Logistic Generalized Linear Model (GLM) with a binomial distribution error, with the species being the predictor factor. Odds ratios (OR) were estimated to quantify the probability of one event relative to another, such as e β being β the model estimator (Agresti, 2015). The odd is defined as the probability that an event occurs divided by the probability it does not. The OR of the variable "species'', for example, for the combination of C. quadrifasciata octodecimpustulata and C. sanguinea, is the ratio of the probabilities of consumption (or non-consumption) of C. quadrifasciata octodecimpustulata divided by the probabilities of consumption (or non-consumption) of C. sanguinea. The analyses were carried out with the statistical package lmer of the R software (R Core Team, 2018 version 3.5.1).
IGP by adults on larvae Synchronous cohorts of neonate larvae of each species were reared until they reached the required instars (L2 and L4). Adult females in the reproductive age of 2-3 weeks old of each coccinellid species were then isolated in the experimental unit with either one L2 heterospecific larva or one L4 heterospecific larva for 24 h, whereupon the frequency of larval mortality was tallied. The frequency of replicates in which IGP occurred was compared between treatments, using Chi-square contingency tests.
IGP between larvae One L4 of each coccinellid species was isolated in the experimental unit with either one L2 or one L4 heterospecific larva for 24 h. After 24 h, the predation events were recorded and used to estimate the intensity and symmetry of IGP. The symmetry index was calculated as the number of replicates in which a given predator was eaten divided by the total number of replicates in which IGP occurred. This value was compared with an expected value of 50% (symmetric interaction) with the Chisquare goodness-of-fit test for each combination of species.

Field sampling
Harmonia axyridis was present throughout the entire crop cycle but it was the most abundant species in the FF stage of the crop. E. connexa and C. sanguinea also were present throughout the entire crop cycle, and they had the highest relative abundance at the beginning and end of the crop cycle, respectively. C. quadrifasciata octodecimpustulata appeared only in vegetative and flowering stages (Fig. 1a), its relative abundance was low and coincided with the highest relative abundance of E. connexa. In general, taking into account all data, the most abundant species was H. axyridis (40.6%), followed by C. sanguinea (37.7%), E. connexa (20.6%), and C. quadrifasciata octodecimpustulata (1.1%).
In relation to the distribution of coccinellids in the plant strata, E. connexa was the most abundant in the lower stratum, H. axyridis and C. sanguinea were present mainly in the upper stratum, and C. quadrifasciata octodecimpustulata was only observed in the lower stratum coinciding also with the highest abundance of E. connexa (Fig. 1b). The number of aphids was higher in the vegetative (V), fruiting (FR) and postharvest (P) phenological stages (H (4, n=4323) = 83.82; P < 0.001; Fig. 1a) and in the lower stratum (H (2, n=4323) = 9.54; P = 0.008; Fig. 1b). Data of all farms and years are given in Tables S1 and S2.
Spatially, at the plant scale, all the coccinellid combinations showed negative association (P(lt) < 0.001) except those that included C. quadrifasciata octodecimpustulata which was not associated with any other species (Table S3). The probability of temporal cooccurrence (at the phenological stage scale) of the different pairs of species was, in general, low (< 25%).

IGP by adults and larvae on eggs
In all combinations tested, adults and larvae predated heterospecific eggs. In general, the intensity of IGP was significantly different between species (Fig. 2; Table S4). Adults and larvae of C. quadrifasciata octodecimpustulata were stronger intraguild predators on eggs of C. sanguinea and E. connexa than vice versa. Similarly, the larvae of C. quadrifasciata octodecimpustulata were stronger intraguild predators on H. axyridis eggs than vice versa, although the predation of adults on eggs was symmetrical between both species (Fig. 2).
The proportion of eggs predated by adults was generally high and significantly different between species. C. quadrifasciata octodecimpustulata preyed a similar proportion of eggs of C. sanguinea and E. connexa and, in turn, these values were higher than vice versa; however, in combination with H. axyridis this trend was reversed (Table 1). Similarly, both L2 and L4 of C. quadrifasciata octodecimpustulata consumed a similar proportion of C. sanguinea and E. connexa eggs, and these were higher than vice versa (Table 2). and Harmonia axyridis (Ha) on heterospecific eggs. *P < 0.05, **P < 0.01, ***P < 0.001, and ns: nonsignificant difference IGP by adults on larvae Adults of all coccinellid species preyed on heterospecific larvae. The intensity of the predation of C. quadrifasciata octodecimpustulata was greater on L2 and L4 of C. sanguinea and E. connexa, than vice versa; whereas H. axyridis was the strongest intraguild predator on C. quadrifasciata octodecimpustulata ( Fig. 3; Table S5).

IGP between larvae
The IGP between L4 of C. quadrifasciata octodecimpustulata and L4 of C. sanguinea and E. connexa, was bidirectional and asymmetric being C. quadrifasciata octodecimpustulata the strongest intraguild predator; however, the interaction was unidirectional between L4 of C. quadrifasciata octodecimpustulata and L4 of H. axyridis being this latter the intraguild predator. The IGP between different larval instar was unidirectional, with the largest larvae always preying on the smaller ones ( Fig. 4; Table S6).

Discussion
At temporal scale, a tendency toward segregation of coccinellid species in eggplant was observed. E. connexa appeared in greater abundance in early stages of the crop, H. axyridis in the middle and C. sanguinea at the end of the crop cycle. The temporal abundance of E. connexa and C. sanguinea was similar to those found by Rocca et al. (2021) in sweet pepper. The low frequency and abundance of C. quadrifasciata octodecimpustulata in relation to the other coccinellid species, recorded in this study, agree with the results obtained by other authors in different crops and wild vegetation (de Souza et al., 2021;Engel et al., 2020;Lixa et al., 2010;Pimenta et al., 2020;Saini, 2004). As observed in other crops in Argentina (Saini, 2004) and other countries (Grez et al., 2016;Martins et al., 2009;Roy et al., 2016), H. axyridis was one of the most abundant species of the coccinellid assemblage, along with C. sanguinea. At spatial scale, a very low probability of cooccurrence of the coccinellid species in eggplant crop was observed. The low abundance of aphids found could affect the coccinellids cooccurrence patterns (Prescott & Andow, 2018) because they are usually aggregated on large aphid colonies (Wright & Laing, 1980;Evans & Youssef, 1992;Elliott & Kieckhefer, 2000). The low prey density could, on the one hand, promote intraguild predation due to lack of extra-guild prey, but on the other hand, by not aggregating in patches of high prey density, the encounter rate would be reduced. The spatial segregation could also be a consequence of coccinellids recognizing heterospecific chemical Fig. 3 Intensity of IGP (frequency of replicates in which IGP occurred) by adults of Coleomegilla quadrifasciata octodecimpustulata (Cq), Cycloneda sanguinea (Cs), Eriopis connexa (Ec) and Harmonia axyridis (Ha) on second (L2) and fourth instar (L4) heterospecific larvae. *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 4
Symmetry index of IGP between heterospecific larvae of the same and different instar of Coleomegilla quadrifasciata octodecimpustulata (Cq), Cycloneda sanguinea (Cs), Eriopis connexa (Ec), Harmonia axyridis (Ha). L2: second instar larvae, L4: fourth instar larvae. *P < 0.05, **P < 0.01, ***P < 0.001 tracks and avoiding laying eggs in sites with such tracks (Agarwala et al., 2003;Michaud & Jyoti, 2007;Růžička & Zemek, 2008), to reduce the probability of participating in IGP (Hindayana et al., 2001;Lucas et al., 1998;Yasuda et al., 2004). The highest number of random associations among coccinellid species found in the flowering stage, would be related to the availability of food resources, such as pollen and prey, in this phenological stage. For example, Musser and Shelton (2003) found that H. axyridis and C. maculata tended to temporarily segregate in corn crops, and the highest cooccurrence was observed in periods in which pollen availability was higher.
In the laboratory, H. axyridis was the strongest intraguild predator in interaction with C. quadrifasciata octodecimpustulata. The same was found in other IGP studies involving H. axyridis-C. sanguinea (Michaud, 2002), and H. axyridis-E. connexa (Ribeiro Pereira dos Santos et al., 2009;Mirande et al., 2015). Moreover, C. quadrifasciata octodecimpustulata in interaction with C. sanguinea and E. connexa was the strongest intraguild predator. It is known that both the eggs and the larvae of some coccinellid species contain alkaloids, toxic substances for other species, avoiding their predation (Daloze et al., 1995;Hautier et al., 2011;Hemptinne et al., 2000). The toxicity of alkaloids to predators and predator ability to tolerate the toxic effect vary greatly between species and individuals of the same species (Agarwala & Dixon, 1992;Cottrell, 2004). For example, larvae and adults of Adalia bipunctata L. were reluctant to eat eggs of Coccinella septempunctata L. (Agarwala & Dixon, 1992). Accordingly, the lower consumption of C. quadrifasciata octodecimpustulata and H. axyridis eggs, by C. sanguinea and E. connexa, could be due to the presence of some toxic compounds (Kajita et al. 2010;Rocca et al. unpublished data). In relation to larvae of the same instar, C. quadrifasciata octodecimpustulata was stronger intraguild predator against C. sanguinea and E. connexa than vice versa, probably due to the presence of deterrent or toxic substances and physical defenses-dorsal spines-of C. quadrifasciata octodecimpustulata (obs. pers.); however, in interaction with H. axyridis, the latter was the strongest intraguild predator. The larvae of H. axyridis are, in general, relatively larger, with physical and chemical defenses, and more aggressive than those of other coccinellids (Michaud, 2002;Yasuda et al., 2004;Kajita, 2010). For example, Anatis ocellata (L.) larvae were little attacked by H. axyridis due to their larger size and the presence of dorsal spines (Ware et al., 2008). Otherwise, in this study it was observed that larger instars preyed on the smallest, regardless of the species, as was recorded for C. septempunctata preying on Hippodamia variegata Goeze larvae (Agarwala & Dixon, 1992;Lucas, 2012).
In general, the intensity of the intraguild predation could be overestimated in experiments in small arenas by confinement in these conditions. Experiments in small arenas could show a high intensity of intraguild predation; however, in field conditions habitat structure would create refugia for intraguild prey, leading to a reduction of the encounters rate and therefore of the intraguild predation, allowing the coexistence of predators (Janssen et al., 2007). Nevertheless, this type of experiment in enclosure conditions provides basic knowledge about the intrinsic characteristics of each species in interaction with other species. Experimental results observed in small enclosures and over small-time frames may, in some cases, still scale up to predict field-wide patterns (Lucas & Rosenheim, 2011). The results obtained in our laboratory study suggest that C. quadrifasciata octodecimpustulata could exclude C. sanguinea and E. connexa from this system. Otherwise, H. axyridis is a strong intraguild predator on C. quadrifasciata octodecimpustulata, as well as on C. sanguinea (Michaud, 2002) and E. connexa (Mirande et al., 2015). In this context, native coccinellid species would be vulnerable to the exotic and invasive H. axyridis; however, the lack of cooccurrence of this species with the native ones, under conditions of low density of extra-guild prey (Agarwala & Dixon, 1992;Lucas et al., 2009;Rocca et al., 2017), would indicate that spatial segregation could be the mechanism by which these species coexist in the eggplant crop. Authors' contributions María Estefanía Rizzo carried out the laboratory assays, performed the analysis of data, interpretation and discussion of results and prepared the draft of the manuscript. Adriana Salvo collaborated in the interpretation and discussion of results, and in the writing of the final version of the manuscript. Margarita Rocca participated in the planning of the assays, collaborated in the analysis of data, interpretation, and discussion of results, and in the writing of the final version of the manuscript. She also obtained the funds to carry out the investigation. Nancy Greco participated in the planning of the assays, interpretation, and discussion of results, as well as in the writing of the final version of the manuscript. She also obtained the funds to carry out the investigation.
Funding This research was supported by grants from the Agencia Nacional de Promoción Científica y Tecnológica PICT 2015-1427, PICT 2019-1745 and by the Universidad Nacional de La Plata Programa de Incentivos N834.

Data availability
We added supplementary material.

Declarations
Ethics approval and consent to participate This manuscript and the authors of the manuscript are not involved in any potential conflicts of interest, including financial interests and relationships and affiliations, and all authors gave their consent to participate in the manuscript.

Consent for publication
All authors gave their consent for the publication of the manuscript.

Competing interests
Authors have no competing interests as defined by Springer, or other interests that might be perceived to influence the results and/or discussion reported in this paper.