Introduction

Food webs are descriptive devices defined by the linkages between consumers and their resources (Kratina et al. 2009). The food web systems are complex and intrinsically dynamic (Kratina et al.2009). In predator-prey interaction, a predator's performance is dependent on numerous characteristics, two of which are the numerical response and functional response (Rahman et al.2012; Juliano 2001). These responses provide data on the biocontrol efficiency of a natural enemy against a particular pest (Rahman et al. 2012; Juliano 2001).

The functional response describes the relationship between an individual's consumption rate and food density (Rahman et al. 2012; Solomon 1949). Three types of functional responses were described by Holling (1959a, 1959b, 1961). The type 1 response is the linear response where the number of prey killed increases linearly to a maximum, and then remains constant as prey density increases. In the type 2 response, the number of prey consumed increases with prey density at a decelerating rate towards an asymptote. A type 3 response, the sigmoid response, depends greatly on the experimental conditions and the feeding history of the predator. While arthropods exhibit both type 2 and 3 responses, type 2 is observed most often (Cedola et al. 2001). The basic factors that affect functional response are: food identification, search rate, capture, prey consumption, and time of exposure of predator to prey (Holling 1959b). Functional response is an important factor in the dynamics of a predator-prey relationship and is a key component of a population model (Sepulveda & Carillo 2008). It has been used to forecast the mechanisms underlying predator-prey behaviour to improve the practical predictive potential of predator candidates for biocontrol. The numerical response is the change in a predator's reproductive output at varying prey densities (Rahman et al.2012; Juliano 2001; Hassell 1978).

The tomato/potato psyllid (TPP), Bactericera cockerelli, is a serious pest of some agricultural and ornamental plants, especially of the Solanaceae family (Butler & Trumble 2012). An outbreak of TPP can result in significant plant damage such as defoliation, stunted growth, and reduced yield (Butler & Trumble 2012). A variety of TPP control methods are utilized around the world; however, most control of TPP populations depends heavily on pesticide application (Mauchline et al. 2013; Walker et al. 2015). Exploitation of natural enemies such as predatory mites is a potential way of minimizing pesticide usage as well as increasing the emphasis on environmentally friendly techniques of pest control.

Predatory mites from the family Phytoseiidae have been important biological control agents in the agriculture industry (McMurtry et al. 2013). These mites have been widely used in biological control programmes of several mites and small insect pests of fruits, vegetables, and other crop species worldwide (Opit et al. 2004; Arthurs et al. 2009; Juan-Blasco et al. 2012; McMurtry et al.2013).

Amblydromalus limonicus and Neoseiulus cucumeris are type III generalist predatory mites and are relatively widely distributed (McMurtry et al. 2013). They can prey on a variety of small arthropods (e.g. whiteflies, scale insects, and mites) and can also feed on other food sources such as pollen across an extensive range of habitats (McMurtry et al. 2013). A. limonicus was first used against spider mites (Oligonychus punicae and Tetrancyhus cinnabarinus) in the USA in the 1960s (McMurtry & Scriven 1971). In the early 1990s, A. limonicus was used successfully to prevent and control the western flower thrips, Frankliniella occidentalis (Messelink et al. 2006). Recently, A. limonicus was commercialised and used in Europe primarily for thrips and whitefly control (Knapp et al. 2013; Medd & GreatRex 2014; Chorazy et al. 2016). In New Zealand, rates of predation of A. limonicus on TPP immature stages were reported at some fixed densities of prey (Xu & Zhang 2015; Davidson et al. 2016; Patel & Zhang 2017). However, there have been no studies determining the functional response of A. limonicus to a range of prey densities. In comparison, one of the top selling biocontrol agents worldwide is N. cucumeris, which is primarily used for controlling small insects such as thrips and phytophagous mites (Van Lenterten 2012). A type 2 functional response has been demonstrated for N. cucumeris against other prey species, e.g. Thrips flavidulus (Yao et al. 2014) and western flower thrips (Shipp & Whitfield 1991). However, no similar studies have been conducted on functional response of N. cucumeris to TPP.

This study is the second of a series of laboratory and greenhouse experiments to compare and evaluate the efficiency of two commercially available species of predatory mites (A. limonicus andN. cucumeris) as potential biocontrol agents of TPP. The first study (Patel & Zhang 2017) examined the preference of TPP eggs over 1^st instar nymphs by A. limonicus and N. cucumeris. The results indicated that N. cucumeris can consume TPP eggs and 1^st instar nymphs and that both predatory mites have the potential to consume more than 10 TPP eggs or 1^st instar nymphs per day. Therefore, this study investigates the functional and numerical responses of A. limonicus and N. cucumeris to TPP. The aims of the study are to examine (a) how predation rates and reproductive rates of A. limonicus and N. cucumeris increase with increasing densities of TPP eggs or 1^st instar nymphs, and (b) if there is a difference in responses to prey density between the two predatory mite species.

Material and Methods

Statisitical analysis

A logistic regression of the proportion of prey consumed as a function of initial prey density was used to determine the shape of the functional repsonse curve of A. limonicus and N. cucumeris to different stages of TTP;

N_e is the number of prey consumed, No is the initial prey density, (N_e/N_o) is the probability of prey consumption and P₀, P₁, P₂,P₃ are the maximum likelihood estimates of the intercept, linear, quadratic and cubic coefficients, respectively (Juliano 2001). To determine the type of functional response, the signs of P₁ and P₂ are used. The predator displays a type 2 response when the linear coefficient is significantly negative (P₁ < 0), which indicates that the proportion of prey consumed declines monotonically with the initial prey density. When the linear coefficient is positive (P₁ > 0), and the quadratic coefficient is negative (P₂ < 0), the predator has a type 3 functional response (Juliano 2001). The logistic regression analysis indicated that our data fit a type 2 functional response for eggs and 1^st star nymphs of TTP for both predatory mite species; therefore, additional analyses conducted were restricted to a type 2 response.

A further analysis was conducted by carrying out non-least squares regression to estimate the attack rate (α) and handling time (Th). The Holling disc model (Holling 1965) was used to model the relationship between numbers of prey consumed (Ne) and initial prey density (No):

N_e is the number of prey consumed, N0 is the initial prey density, α is attack rate, T is total predation time (1d = 24 hrs), and T_h is the handling time. Parameters with non-overlapping 95% confidence intervals were considered significantly different when parameters were compared across treatments.

All the analyses were performed using R Studio. The numbers of predator eggs laid on an egg or 1^st instar diet across densities were compared using a two-way ANOVA. Analyses were carried out for both raw and log-transformed responsive variables and both displayed similar results. The results presented are from raw data only.

Results

Functional response

The interaction between predator species, density and prey stage was significant for the number of prey eaten (F=4.024; df=22; P=0.000). Thus the effects of predator species and prey density were analysed separately for eggs or 1^st instar nymphs of TPP.

The interaction between predator species and TPP egg density was significant for the number of eggs eaten (F= 2.968; df=7; P=0.006), indicating different curves between the two species. The two curves overlapped initially at very low densities when both species could consume most of the prey provided, diverged gradually until the density of 16, and then diverged again after a short period of plateau between densities of 16 to 20 (Fig. 1). There was also a significant effect of density (F=19.047; df=7; P=0.000) and species (F=86.381; df=1; P=0.000) of the number of eggs eaten. A. limonicus consumed approximately twice the amount of eggs as did N. cucumeris across all densities except the lowest density of 2 (Table 1).

The interaction between predator species and prey density when the species were fed on 1^st instar nymphs was significant for the number of prey eaten (F=5.763; df=7; P=0.000), indicating different curves between N. cucumeris and A. limonicus. The curves overlapped initially at very low densities when both species did eat most of the prey, diverged gradually until the density of 16, and then reached a plateau (Fig. 2). There was also a significant effect of density (28.698; df= 7; P=0.000) and species (158.975; df= 1; P=0.000) on the number of 1^st instar nymphs eaten. At densities higher than 4, A. limonicus consumed more than twice the number of 1^st instar nymphs than did N. cucumeris(Table 1).

TABLE 1.

The number of prey consumed in 24 hours by each female of A. limonicus and N. cucumeris when presented with densities of 2, 4, 8, 16, 20, 24, or 28 eggs or 1^st instar nymphs of TPP. Data in the format of mean ± s.e. (range). The number of replicates is 10 for each treatment.

TABLE 2.

Maximum likelihood estimates from logistic regressions of the proportion of prey consumed as a function of initial TPP prey density by adult female A. limonicus and N. cucumeris.

In the absence of predators, egg and 1^st instar mortality ranged between 1% and 15%, thus reflecting very little natural or manipulation mortality; therefore data for functional response did not have to be corrected.

Both predatory species displayed a type 2 functional response when fed on TPP eggs or 1^st instar nymphs, as revealed by significant negative values of P₁ (Table 2).

FIGURE 1.

Functional response of adult female A. limonicus (upper line and round dots) and N. cucumeris(lower line and triangular dots) fed on TPP eggs. Points represent the observed number of prey consumed at each initial prey density. Grey areas around the curves represent 95% confidence intervals.

Handling time and attack rate

Amblydromalus limonicus had a higher attack rate than N. cucumeris for both eggs (>1.6 times) and 1^st instar nymphs (3 times) (Table 3). However, N. cucumeris had a longer handling time than A. limonicus for both eggs (>twice) and 1^st instar nymphs (>twice) (Table 3).

FIGURE 2.

Functional response of adult female A. limonicus (upper line and round dots) and N. cucumeris(lower line and triangular dots) fed on TPP 1^st instar nymphs. Points represent the observed number of prey consumed at each initial prey density. Grey areas around the curves represent 95% confidence intervals.

Reproductive rates

The interaction between predator species and density on an egg diet was not significant for the number of eggs laid (F=1.437; df=7; P=0.195). While density had no significant effect on the reproductive rate (F=1.543; df=7; P=0.157), predator species did have an effect on the number of eggs laid (F=29.569; df=1; P=0.000). A. limonicus had a much higher reproductive rate than N. cucumeris across all densities when fed on eggs (Table 4).

The interaction between predator species and density when fed on 1^st instar nymphs was significant for the number of eggs laid, indicating that the two species showed different responses to prey density (F=2.772; df=7; P=0.010; Table 4). There was no significant density effect on the reproductive rate (F=1.207; df=7; P=0.302); however, there was an effect of predator species on the number of eggs laid (F=87.309; df=1; P=0.000). A. limonicus had a significantly higher reproductive rate than N. cucumeris across all densities when fed on 1^st instar nymphs (Table 4).

TABLE 3.

Mean values ± SE of attack rate and handling time (hours) for A. limonicus and N. cucumeris feeding on different life stages of TPP.

TABLE 4.

The number of eggs laid by A. limonicus and N. cucumeris at each density when fed on an diet of TPP eggs or 1^st instar nymphs. Data are presented in the format of mean ± s.e. The number of replicates is 10 for each density.

Discussion

In this study, we compared the functional responses of A. limonicus and N. cucumeris to the density of TPP eggs or 1^st instar nymphs. The negative values estimated for the linear parameters as well as the inverse density-dependant relationship between the proportion of prey consumed and initial prey density of eggs and 1^st star nymphs indicate a type 2 functional response for both A. limonicus andN. cucumeris. In several phytoseiid species, type 2 functional response is common, including N. cucumeris on western flower thrips (Shipp & Whitfield 1991), N. cucumeris, Neoseiulus barkeri, andEuseius nicholsi on Thrips flavidulus (Yao et al. 2014), N. cucumeris on Thrips tabaci larvae (Madadi et al. 2007), and N. cucumeris on T. tabaci (Nielsen et al. 2014). Predators with a type 2 functional response have proved to be efficient, especially at low prey densities (Koehler 1999). Patel & Zhang (2017) estimated the predation rates of TPP eggs and 1^st instar nymphs by A. limonicus and N. cucumeris at the fixed density of 10 and the results are in line with our results in this paper. For example, A. limonicus consumed on average 6.17 eggs when offered 10 TPP eggs in Patel & Zhang (2017), and in this study, A. limonicus consumed 5.70 TPP eggs when offered 8 and 7.00 TPP eggs when offered 12.

Predation rates for both species increased with increasing prey density, perhaps due to interference or disturbance by the prey. This has been noticed in Amblyseius largoensis, Euseius corcordis, and Galendromus (=Typhlodromus) helveolus (Sandness & McMurtry 1970). At higher prey densities, the amount of time spent on consuming individual prey may be reduced because prey may accidently bump into a feeding predator, which could cause the predator to abandon the prey it is eating and attack another prey. The bumping of prey into each other (only for first instar nymphs of TPP here) could cause the prey to move, therefore increasing its chances of encountering a predator. This higher prey density causes wasteful killing (Metz et al. 1988) as a predator may partially consume the prey instead of eating one whole prey.

Even though it is common to have a type 2 functional response among phytoseiid species, functional responses may change due to a number of factors. In our study, N. cucumeris displayed a type 2 functional response when fed on TPP eggs and 1^st instar nymphs; however, when Fang et al.(2013) compared the effectiveness of N. cucumeris and N. barkeri in controlling the citrus psyllid (Diaphorina citri) in China, the functional response of N. cucumeris was type 3 when fed D. citrieggs. Reaction to environmental conditions may cause a type 2 response shifting to a type 3 response. Functional response may be influenced by different factors, e.g. abiotic, such as temperature and relative humidity, and biotic, such as prey or host species (Donnelly & Phillips 2001; Hoddle 2003; Allahyari et al. 2004; Faria et al. 2004); the presence of alternative food (Wei & Walde 1997); predator sex (Parajulee et al. 1994) and predator age and feeding history (Eveleigh & Chant 1981). In addition, the host plant may also influence functional response, for instance indirectly, by affecting prey palatability or offering prey refuges (Messina & Hanks 1998) or directly, by plant structure such as trichome or by sticky exudates that reduce searching efficiency (Cedola et al. 2001). Holling (1965) stated that predators that have a type 3 functional response contribute to regulating their prey populations because this type of response allows for long-term persistence.

Attack rate and handling time are parameters that also help determine the magnitude of functional responses (Omkar 2005). The handling time is an important and reliable indicator of the amount of prey consumed and the predator efficiency as it determines the cumulative time spent on capturing, killing, and digesting the prey (Veeravel & Baskaran 1997). A. limonicus had an attack rate of 0.05 for eggs and 0.06 for 1^st instar nymphs, whereas N. cucumeris had an attack rate of 0.03 for eggs and 0.02 for 1^st instar nymphs. The attack rates of two prey stages were similar within each predator species, which could be due to the TPP eggs and 1^st instar nymphs being similar in size. However, there was a significant difference between predator species: A. limonicus had a higher attack rate for both prey stages of TPP than N. cucumeris had. This implies A. limonicus could be a more effective control agent in early infestation when the prey population is comprised of more TPP eggs and 1^st instar nymphal stage.

Furthermore, A. limonicus had a handling time of 1.74 hrs for eggs and 1.15 hrs for 1^st instar nymphs, whereas N. cucumeris had a handling time of 3.68 hrs for eggs and 2.50 hrs for 1^st instar nymphs. For both species of predatory mites, handling time for eggs was higher than for 1^st instar nymphs. Various factors may affect handling time such as predator speed, movement of prey, and the time spent subduing individual prey (Hassell 1978), which could possibly be related to behavioural and physical prey attributes (Ali et al. 2011). Madadi et al. (2007) and Yao et al. (2014) also demonstrated similar results to our study in regards to attack rate and handling time as both studies investigated the functional response of N. cucumeris on Thrips tabaci larvae and first instar of T. flavidulus. Madadi et al. (2007) indicated the attack rate for N. cucumeris was 0.048 and the handling time was 1.030 hrs at 25°C. Yao et al. (2014) indicated the attack rate for N. cucumeris was 0.8856 and the handling time was 0.0298 hrs at 26°C. Making direct comparisons of these estimates by these authors with our results is, however, problematic due to differences in prey, environmental conditions, and prior predator adaptation to the tested prey.

Unlike the functional response, the numerical response of both species did not increase with prey density. There was no apparent trend of reproductive rates across densities for both TPP eggs and 1^st instar nymphs. However, A. limonicus consistently produced significantly more eggs than did N. cucumeris across all densities when fed on an diet of either eggs or 1^st instar nymph of TPP. The implications of a non-significant numerical response are that both predatory mite species would be unable to keep up with the growth of a TPP population if infestation should occur. Therefore, as suggested previously by Xu & Zhang (2015), inoculative releases would be required to prevent pest development in the early season; this may be achieved by the release of biocontrol agents to allow reproduction on alternative food such as pollen to occur before prey build-up. This would enable the setting up of a longer term control by keeping TPP populations to low levels—prevention rather than a curative control.

Nevertheless, functional response experiments carried out under laboratory conditions may have limited value for determining characteristics of predation in a natural setting. Some of the criticisms are that: the small arenas used to carry out the experiments may not be representative of the natural searching ability of a predator; environmental complexity cannot be easily replicated through a simple experimental arena; and plant characteristics can also affect predation response of a predator (Messina & Hanks 1998; O'Neil 1989). Great caution is therefore needed in extrapolating the results of oversimplified laboratory experiments to more complex and heterogeneous field conditions. Future experiments, being carried out on greenhouse potted plants to show more realistic predator-prey interaction, will be reported in a future paper.