Frequency and quantity of alternative food supply
influences the effectiveness of biological pest control
Annabel LandmanResearch internship carried out by Hanneke de Lange and Annabel Landman
Supervised by Arne Janssen, Institute of Biodiversity and Ecosystem Dynamics (IBED), University of Amsterdam
29/06/2020
Abstract
Natural enemies are used for biological pest control, to reduce the population densities and distribution of many pest species. Natural enemies are often generalists, feeding on different food sources, forming a complex food web with indirect interactions between individuals and populations. By supplying alternative food, such as pollen, to generalist predators, their populations can be established and maintained in the crop, even in absence of the pest, thus working as preventive crop protection. However, there is a lack of research regarding what the optimal frequency or quantity of supplying alternative food is for biological control. What is needed are natural enemy densities high enough to control pests, and their densities can be increased with alternative food. However, feeding too much and too frequently may result in temporal satiation of the predators, thus offering opportunities for pests to invade and establish. To study this dilemma, a predator-pest-pollen model was constructed to research what the optimal frequency and quantity is of alternative food supply for a high and low initial predator density. Several frequencies and intervals were tested from which the maximum density of prey larvae, and the mean prey larvae, prey and predator densities were calculated.
Simulations show that the highest prey and prey larvae densities occurred with a low initial predator density and with short intervals of supplying alternative food. In contrast, with a higher initial predator density, the highest prey and prey larvae densities occurred with long intervals. Hence, it can be concluded that supplying alternative food to predators will sometimes result in less optimal pest control. There is an optimum frequency and quantity of alternative food supply; more frequently or higher quantities are not always better and there is no clear strategy that works for all cases but it depends on factors such as the initial densities. Although, supplying alternative food to natural enemies is a good strategy for pest control, the timing and quantities need to be fine-tuned to the specific pest-natural enemy system.
Keywords: biocontrol, natural enemies, generalist predator, apparent competition, apparent mutualism, population dynamics, alternative food, resource pulses and population model
Table of contents
1. Introduction 3
2. Methods
2.1 Predator-prey-pollen model 4
2.2 Simulation of the model 6
3. Results
3.1 Time series of the predator-prey-pollen model 8
3.2 Predator-pollen dynamics 10
3.3 Predator-prey-pollen dynamics 10
4. Discussion
4.1 Perspectives for biocontrol 13
4.2 Suggestions for future research 14
5. Acknowledgements 15
a
b
1. Introduction
Since the last decades of the twentieth century, the demand for biological pest control in greenhouses has been increasing due to the negative impact of pesticides on the environment (van Lenteren et al., 2018). Pesticides negatively affect biodiversity, the behaviour of species, aquatic food webs, and human health (Crall et al., 2018; Yamamuro et al., 2019). In contrast with pesticides, biological control is a sustainable method using populations of other organisms (natural enemies, pathogens) to reduce densities of the pest species (Crall et al., 2018; van Lenteren et al., 2018). The ‘Sustainable Use of Pesticide Directive’ of the European Union (EU) has been supporting the use of biological pest control. Besides, policymakers, consumers, politicians, farmers, and many more urge for a decrease in pesticide use (van Lenteren et al., 2018).
One of the methods of biological pest control is the use of natural enemies (predators, herbivores, omnivores, etc.) because they can reduce the population density and distribution of several pest species (Bale et al., 2008). For decades, specialist predators, which feed on one specific pest, have been used in biocontrol (Janssen & Sabelis, 2015). Nowadays, natural enemies are often generalist predators, feeding on several food sources, forming part of complex food webs with indirect interactions among pest and predator species (Janssen et al., 1998; Messelink et al., 2008; Nomikou et al., 2010). Holt (1977) developed a theory that addresses one of these interactions between prey species sharing a natural enemy. He stated that this mechanism leads to apparent mutualism or apparent competition (Figure 1).
The first indirect interaction, so-called apparent mutualism (Figure 1a), refers to a positive interaction between prey that share a predator, which occurs in the short-term when the predation of
predators is limited due to saturation or switching (non-linear functional response type II) (Holt & Bonsall, 2017; Nomikou et al., 2010). Increased densities of one prey species will result in saturation of the predator population, resulting in a decrease of predation of the second prey species. In contrast, apparent competition (Figure 1b) indicates a negative interaction between prey in the long-term (Nomikou et al., 2010). This occurs when the density of one prey species increases and benefits the predator’s abundance, which in turn harms both prey populations (Holt & Bonsall, 2017).
Figure 1. Schematic overview: a. Apparent mutualism; two prey (R1 and R2) share a predator (C), which leads to
a positive interaction in the short-term, due to a saturating functional response of the predator and b. apparent competition leading to a negative interaction in the long-term (Holt & Bonsall, 2017).
Since Holt (1997) introduced the theory behind these indirect interactions, more research has been done regarding this subject. Messelink et al. (Messelink et al., 2008) concluded that the predator’s density was significantly higher in the presence of two pest species as opposed to one pest species, resulting in better pest control due to apparent competition. Besides, the mixed diet of two pest species had a positive effect on the juvenile predator survival and development rate. One method of biological pest control with generalist predators is to add alternative food, such as pollen, to invoke apparent competition between the alternative food and the pest (Rijn et al., 2002). By supplying alternative food, generalist predators can establish
and persist even in absence of the pest (Rijn et al., 2002). Because food is abundant, the predator’s mortality rate decreases and/or its population growth rate increases and total predation increases, causing the pest density to decrease.
Apparent competition plays an important role in biological pest control with generalist predators (Janssen & Sabelis, 2015), and providing alternative food for natural enemies is already proven to be successful (Holt & Bonsall, 2017; Rijn et al., 2002). However, there is a lack of research regarding the optimal frequency and quantity of supplying alternative food, resulting in so-called resource pulses, for biological control. Resource pulses can change community structures and prevent coexistence in stable environments or enable coexistence when it would naturally not occur (Holt, 2008; Sato et al., 2016). Holt (2008) researched predator-prey dynamics with different resource subsidies, which were eaten by the prey. He concluded that long resource pulses, as opposed to short pulses with the same quantity, induced more instability of the dynamics of the system. Besides this, a high abundance of food for predators due to resource pulses may cause temporal reductions of predation on the prey species due to saturation. In contrast, predator densities may not be sufficient to control pest densities when the pulse quantity is too low.
Therefore, to prevent crop damage, it is of great importance to find out what the optimal frequency and quantity of resource pulses are in a predator-prey-alternative food model with apparent competition. In the short-term, it could give new insights regarding resource subsidies in a predator-prey-pollen system and could be used as a background for practical research later on. In the long-term, a decrease in crop damage will have an economic advantage for agriculture, especially for agriculture in greenhouses, where alternative food is
used increasingly nowadays (Messelink et al., 2014). The obvious economic advantages of less frequent supply of alternative food are lower costs (material and labour), but this may impede good control.
For those reasons the specific research questions of this project are: ‘What are the short-
and long-term effects of resource pulses in a predator-prey-pollen system?’, ‘What is the optimal interval between resource pulses to maintain the pest density under control?’ and ‘What is the optimal quantity of the resource to maintain the pest density under control?’. To provide an answer, a theoretical
research was performed using a predator-prey-pollen-model, and several frequencies and quantities of resource subsidies were tested.
2. Methods
2.1 Predator-prey-pollen model
To determine the optimal interval and quantity of resource pulses in a prey-predator system with alternative food and apparent competition between this food and the pest, the stage-structured predator-prey model of van Rijn et al. (2002) was used and adjusted. van Rijn et al. (2002) researched how plants can benefit from alternative food for predators, even when it is edible for the pest. He constructed a predator-prey-pollen model with differential equations with several assumptions (van Rijn et al., 2002). Firstly, a constant amount of pollen grains was added in the greenhouse per unit of time and pollen density decreased due to natural decay and predation by pest and predators. In addition, the pollen grains were applied concentrated at particular plant parts.
Secondly, the pest population was structured in three different life stages, which entail a juvenile phase that is vulnerable to predation (small larvae), an invulnerable juvenile phase (large larvae), and an invulnerable reproduction phase
(fertile adult female) (Rijn et al., 2002). They assumed unlimited growth of the pest population because the carrying capacity would (hopefully) not be reached in a greenhouse. Besides this, the reproduction rate of thrips feeding on pollen was double that of thrips feeding on plants only. Furthermore, the small larvae suffered from predation with a satiating (type two) functional response.
Lastly, the stage-structured predator population of van Rijn et al. (2002) entails a nonfeeding juvenile phase (eggs and larvae), a feeding juvenile stage (nymphs and non-ovipositing females) and an adult phase (ovipositing females). They assumed that predators do not have a preference for pollen or the pest, but in the presence of pollen, the natural enemies prey less on the thrips due to satiation. Another assumption was that the reproduction rate was affected by the pollen and prey densities, meaning that less food would result in fewer offspring. Besides this, van Rijn et al. (2002) considered that the predator’s mortality increased with lower food availability.
Some changes were made to the stage-structured predator-prey model of van Rijn et al. (2002) to fit the model to the aim of this research. The main distinction was that here the alternative food was exclusively consumed by natural enemies and not by the herbivore. This was done to make the model more general because the feeding of a pest on the alternative food of the predator is rather particular for thrips. Therefore, this aspect was changed in the differential equations of the pollen and prey. Another aspect was that van Rijn et al. (2002) assumed a constant rate of pollen production; this was changed to fluctuating resource pulses. Moreover, the pollen distribution was uniform in this research, which also differs from van Rijn et al. (2002). Finally, van Rijn et al. (2002) assumed infinite growth of the prey population. This could be
a problem when looking at the dynamics over a longer time period because some schedules of pollen provisioning could result in infinite growth of the thrips population.
The adjusted predator-prey-pollen model of van Rijn et al. (2002) consists of 7 differential equations. Firstly, the pollen grains are added in pulses, indicated with pp in the differential equation. The quality of the pollen density decreases with a constant rate of natural decay (b), and pollen are consumed by predators with a functional response
FA(N1, A).
𝑑𝐴
𝑑𝑡 = 𝑝𝑝 − 𝑏𝐴 − 𝐹!(𝑁", 𝐴)𝑃#
Pollen grains are consumed by the juvenile (P2) and
adult (P3) predators (together PC), with juveniles
consuming a fraction j of what adults consume: 𝑃# = 𝑗𝑃$+ 𝑃%
Pollen consumption by predators follows a saturating type II functional response:
𝐹!(𝑁", 𝐴) =
F𝑓!𝐴
𝑁"+ 𝑁&+F𝐴 + 𝑘𝐴𝑁"
fA is the maximum consumption rate of pollen by
predators. Besides, the consumption of pollen depends on the density of the prey (N1), with half
saturation prey density Nf, k the reduction of
predation of the pest because of the presence of pollen, and Φ the food value of pollen relative to the prey.
Secondly, the stage-structured thrips population is: 𝑑𝑁"
𝑑𝑡 = 𝑁%𝑅(𝐴) − 𝐹'(𝑁", 𝐴)𝑃#− 𝑑"𝑁" 𝑑𝑁$
𝑑𝑁%
𝑑𝑡 = 𝑑$𝑁$− 𝑣𝑁%
The vulnerable phase of the prey population (N1)
develops into the invulnerable phase (N2) with a per
capita rate d1. In addition, this vulnerable phase is
consumed by predators (PC), following a saturating
type II functional response: 𝐹'(𝑁", 𝐴) =
𝑓'𝑁"
𝑁"+ 𝑁&+F𝐴 + 𝑘𝐴𝑁"
This functional response is similar to the saturating response of pollen consumption, except that it contains the maximum predation rate on prey (fN)
instead of a maximum consumption rate of pollen (fA). The vulnerable prey population increases with
the net per capita reproduction of fertile females (N3), following a Michaelis-Menten function:
𝑅(𝐴) = 𝑟 ∙ 𝑚𝑎𝑥 :0, 1 −𝑁"+ 𝑁$+ 𝑁% 𝐶 >
r is the maximum reproduction rate of the adult pest
females and C is the carrying capacity, which prevents unlimited growth. The invulnerable phase (N2) increases with the development of the
vulnerable phase (N1) and develops with rate d2 into
adults (N3). Besides, the adult phase has a per capita
mortality rate of v.
Lastly, the predator population is structured in three life stages as well:
𝑑𝑃" 𝑑𝑡 = G(𝑁", A)𝑃%− 𝑒1𝑃" 𝑑𝑃$ 𝑑𝑡 = 𝑒"𝑃"− 𝑒$𝑃$ 𝑑𝑃% 𝑑𝑡 = 𝑒$𝑃$− 𝜇(𝑁", A)𝑃%
The non-feeding juveniles (P1) develop into feeding
juveniles (P2) with rate e1, and are reproduced by
adult predators (P3), following a Michaelis-Menten
function:
G(𝑁", A) = max (0, 𝑔 :''!(F!
!(F!('"− 𝑚>)
The reproduction of adult predators has a maximum reproduction rate g, and is limited by maintenance costs (m), which includes all energy that is required to survive without reproducing. Besides, reproduction depends on the density of the prey (N1),
with half saturation prey density Ng, and Φ the food
value of pollen relative to the prey. The feeding juveniles develop into adult predators with a per capita rate e2. The last differential equation of the
model is of the adult predators, which die with the following equation:
𝜇(𝑁", A) = min (𝜇*, 𝜇 :
𝑁"+F𝐴 + 𝑁+
𝑁"+F𝐴 >)
This inverse of the Michaelis-Menten function indicates that the per capita mortality rate increases with decreasing food availability where μ0 is the
maximum mortality rate at low food densities and μ at high food densities.
The specific parameter values used during the simulations are listed in Table 1 and are based on estimates of the predator-prey system studied by van Rijn et al. (2002) (A. Janssen, pers. comm.). The initial densities for the prey and predator are: N1 =
0.168, N2 = 0.39, N3 = 0.042, P1 = 0, P2 = 0 and P3 =
0.1 or 5 (van Rijn et al., 2002; A. Janssen, pers. comm.).
2.2 Simulation of the predator-prey-pollen model
Before the simulations in R (R Core Team, 2019) started, thresholds for the vulnerable and invulnerable juvenile phases of the prey population (N1 + N2) were determined as 30 prey per plant (van
Rijn et al., 2002; A. Janssen, pers. comm). Above this threshold density, the pest will cause significant economic damage. Moreover, it was important to investigate what happened during shorter periods rather than looking at the equilibrium states of the model.
Table 1. Default parameter values used in the predator-prey-alternative food model (van Rijn et al., 2002; A. Janssen, pers. comm.).
Parameter Description Value Unit
Pollen dynamics
pp Pulsed subsidy of pollen varying pollen grains * plant
pulse time Time between each pulse varying d-1
b Natural decay rate 0.21 * 0.8 pollen grains * d-1
j Fraction of juvenile consumption relative to adult
predator consumption rate
0.25 ratio
Prey dynamics
d1 Per capita developmental rate of vulnerable phase
into invulnerable immatures
1/3 d-1
d2 Per capita developmental rate of invulnerable
immatures into adult
1/15 d-1
v Per capita natural mortality rate of adult prey 0.11 d-1
Predator dynamics
e1 Per capita developmental rate of eggs into juveniles 1/3 d-1
e2 Per capita developmental rate of juveniles into adults 1/5 d-1
Functional responses
Φ Food value of pollen relative to prey 0.34 prey * 104 pollen grains
fA Maximum consumption rate of pollen by de
predators
0.085 pollen grains * adult-1 * day-1
Nf Half-saturation density of vulnerable prey 1.5 prey * plant
k Reduction in predation of the prey because of
presence of pollen
0.04 plant * 10-4 pollen grains
fN Maximum rate of predation on prey 4 prey * adult-1 * d-1
Numerical responses
C Constant determining the carrying capacity of the
prey
105.82
r Maximum reproduction rate of the prey adults in
absence of maintenance costs
2.0 offspring * adult-1 * d-1
g Maximum rate of reproduction of the adult predators
in absence of maintenance costs 𝑔 =
1.5 (1 − 𝑚)
offspring * adult-2 * d-1
Ng Half-saturation prey density 1.0 prey * plant
m Maintenance costs (relative to the total maintenance
and reproduction)
0.2 ratio
μ0 Maximum adult predator mortality rate at very low
densities
0.2 d-1
μ Minimum adult predator mortality rate at very high
densities
0.0625 d-1
Nμ Prey density for which the mortality is half its
maximum
This was because cropping systems in greenhouses do not persist for long enough to allow reaching an equilibrium. To test if a higher initial predator density is beneficial in decreasing crop damage, two different initial values of the adult predator (P3) were
tested: 0.1 and 5 predators/plant.
Both the optimal quantity and interval of pollen supply were determined. As a starting point, 0.17 x 104 pollen grains per plant were used (van Rijn et al., 2002; A. Janssen, pers. comm.). To determine the optimal quantity, the resource pulse was simulated from 0 to 3.4 x 104 pollen grains per plant with steps of 0.1 x 104 pollen grains. The maximum of 3.4 x 104 pollen per plant was chosen because it is 20 times the starting point, which is a sufficient amount of pollen to test. Besides, the interval of pollen supply varied from 1 to 31 days with steps of 3 days. For each tested interval, the maximum values of the juvenile prey density (N1 and
N2) per simulation were plotted against the pulse
quantity to check if the pest threshold density was not exceeded. Moreover, crop damage was estimated by using the mean prey- and prey larvae densities: higher mean or cumulative density means more crop damage. In the subsequent, the mean values of the prey, juvenile prey and predator densities are plotted against each tested resource pulse for various intervals.
3. Results
3.1 Time series of the predator-prey-pollen model
The predator-prey-pollen dynamics over a period of 100 days is shown in Figure 2, with time in days on the x-axes and density in numbers/plant on the y-axis. The pollen was supplied at 0.1 x 104 pollen grains/plant every 31 days (Figure 2a) and decreases because of consumption by the predator and natural decay. Firstly, the stage-structured prey population
increases in time due to a relatively low density of feeding predators and subsequently decreases because vulnerable prey juveniles are consumed by the increased feeding predator stages (Figure 2b). The increase has a developmental time delay: the vulnerable juveniles peak occurs first, then the invulnerable juveniles, and then the adults.
The predator density increases due to the consumption of pollen and prey, and decreases with decreasing prey and pollen densities (Figure 2). Besides, the predator densities increase with a developmental time delay (Figure 2). The predator density increases a second time between 50 and 70 days (Figure 2c). The increase around 50 days is caused by the small increase of vulnerable prey juveniles, which could increase due to a lower predator density. Moreover, the predator density increases further after 62 days because of a new resource pulse.
To conclude, the total prey densities increase when the total predator densities are low and decrease with increasing total predator densities (Figure 2d). The total predator densities increase with increasing total prey densities and with pollen supply, and decrease with a decrease in pollen and prey densities.
The model was also run with a pulse of 1.5 x 104 pollen grains/plant and an interval of 31 days (Figure 3a). The prey dynamics show the same pattern as in the previous simulation; first it increased due to a low density of feeding predators and subsequently the density decreased due to predation (Figure 3b & c). After each pulse, the vulnerable prey phase increased because predators became temporarily satiated by feeding on pollen, resulting in a decrease of predation on vulnerable prey (apparent mutualism). Furthermore, the predator densities increased faster with higher pulse quantities which resulted in an increased consumption of prey.
Figure 2. Time series of 100 days (horizontal axis) of the predator-prey pollen model with a pulse quantity of 0.1 x 104 pollen
grains/plant, and an interval of 31 days. (a) Represents the pollen dynamics, (b) the dynamics of each prey stage, (c) the dynamics of the predator stages, and (d) the dynamics of the total prey, total predator and pollen.
Figure 3. Time series of 100 days (horizontal axis) of the predator-prey pollen model with a pulse quantity of 1.5 x 104 pollen
grains/plant, and an interval of 31 days. (a) Represents the pollen dynamics, (b) the dynamics of each prey stage, (c) the dynamics of the predator stages, and (d) the dynamics of total prey, total predator and pollen.
The peak of the pest population was lower than with a pulse of 0.1 x 104 pollen grains/plant (Figure 2), and therefore, the predator population could not grow as much. The predator density increased again after each pollen pulse, around 60 and 90 days (Figure 3c).
To conclude, the total prey population increased when the predation by feeding predators lowered which occurred after the pollen pulses (Figure 2d). The total predator population increased with increasing prey densities and after resource pulses.
3.2 Predator-pollen dynamics
Figure 4 is a summary of many different simulations with pulse intervalsfrom 1 to 31 days with steps of 3 days and pulse quantities from 0 to 3.4 x 104 pollen grains/plant with steps of 0.1 x 104 pollen grains/plant. Each simulation resulted in data as in Figures 2 & 3, but were summarized by calculating the mean and maximum densities of the pest and predators (hence, each panel in Figure 4 is the result of 374 simulations). The model was run with only the predator population and pollen pulses of various quantities and intervals to determine the effect of pollen pulses on predator dynamics (Figure 4). When the initial density of adult predators was 0.1
predators/plant, the mean predator density increased faster with increasing pulse quantity and with shorter intervals (Figure 4a). The same occurs with an initial density of 5 adult predators/plant, but the mean density is higher for each interval (Figure 4b). Thus, in the absence of the prey, the predator population increases with a more frequent supply and a higher pulse quantity of alternative food.
3.3 Predator-prey-pollen dynamics
Furthermore, the predator-prey-pollen dynamics were simulated using an initial value of 0.1 adult predators/plant (Figure 5). The maximum prey larvae density was plotted against the pulse quantity to check if the threshold value of 30 larvae/plant was exceeded (Figure 5a). For each interval, the larvae density was below the threshold value between a pulse quantity of 0.8 x 104 to 3.4 x 104 pollen grains/plant. Therefore, pulses lower than 0.8 x 104 pollen/plant were not further taken into account because exceeding the threshold would cause significant crop damage and might cause growers to use chemical control.
The mean prey larvae and total prey densities decreased with increasing pulse quantity for intervals of 4 to 31 days, which is the net effect of apparent competition (Figure 5b & 5c).
Figure 4. The effect of varying pulse quantities and intervals on the predator-prey-alternative food model without prey population over a period of 100 days. This figure is a summary of many different simulations, per interval (1 to 31 days) and per pulse quantity (0 to 3.4 x 104 pollen grains/plant) the mean predator density was calculated. (a) Shows the mean predator
densities with an initial density of 0.1 adult predators/plant and (b) represents the mean predator densities with an initial density of 5 adult predators/plant.
However, there may be periods of apparent mutualism where the prey larvae population could increase due to satiation of the predator by pollen pulses (Figure 3). In contrast, the prey densities increased with a pollen supply interval of one day and increasing pollen quantities. This short interval and high pollen supply rate results in so much pollen that the predators remain satiated for a significant part of the total interaction period, resulting in the temporal escape of thrips larvae (apparent mutualism).
Lastly, the mean predator density increased with increasing pulse quantity, and it decreased with longer intervals (Figure 5d). For intervals from 13 to
31 days, the predator density decreased with increasing pulse quantity because a higher pollen pulse caused the predator population to increase faster, which lowers the prey population, causing the predator to subsequently increase less (Figure 5d). This effect can also be seen in Figure 2 and 3. A one-day interval caused the predator density to quickly increase for each interval because the net reproduction rate increases with pollen abundance. With an initial density of 0.1 adult predators/plant, the lowest mean prey density was found at a high pulse quantity and an interval longer than one day because the overall effect of pollen on pest densities is negative.
Figure 5. The effect of varying pulse quantities and intervals on predator and prey densities in a predator-prey-alternative food model with an initial value of 0.1 adult predators/plant (P3). This figure is a summary of many different simulations, per
interval (1 to 31 days) and per pulse quantity (0 to 3.4 x 104 pollen grains/plant) the maximum prey larvae density, and the
mean prey, prey larvae and predator densities were calculated. (a) Shows the maximum prey larvae densities, (b) represents the mean prey larvae densities, (c) the mean prey densities, and (d) the mean predator densities.
The predator-prey-pollen model was also simulated with an initial value of 5 adult predators/plant to determine if crop damage could be prevented even more by releasing higher numbers of predators (Figure 6). The prey larvae density never exceeded the threshold value of 30 larvae/plant (Figure 6a). Besides, the maximum larvae density increased with longer intervals.
The mean total prey and prey larvae densities were lower with the release of higher densities of predators because a higher predator density could consume more prey and was satiated less fast (Figure 5 & 6). Therefore, the negative effects of the pollen supply on the pest species are stronger than the positive effects. Besides, the mean prey and prey larvae densities decreased with increasing pulse quantity, and they decreased more gradually with increasing interval. An increase in
pulse quantity caused the predator density to increase, which in return resulted in the consumption of more prey. Besides, the mean prey and larvae densities are the highest with a long interval and low pollen supply, because there is less food available for the predator population which decreases its density. The fixed simulation length of 100 days cuts off the dynamics at a random phase of the fluctuations, causing fluctuations in the curves at longer intervals (Figures 6a, b & c).
Lastly, for an interval of one day, the mean predator density increased with increasing pulse quantity, because of a higher net reproduction rate (Figure 6d). The predator density decreased with increasing interval length because of a lower pollen supply, which means less food. The lowest crop damage would occur when a high pulse quantity is combined with a short interval.
Figure 6. The effect of varying pulse quantities and intervals on predator and prey densities in a predator-prey-alternative food model with an initial value of 5 adult predators/plant (P3). See legend to Figure 5 for further explanation.
4. Discussion
4.1 Perspectives for biocontrol
For a long time, supplying alternative food has been advocated in biocontrol to reduce crop damage (Holt & Bonsall, 2017; Rijn et al., 2002). However, the optimal frequency and quantity of alternative food supply have not been researched yet. This study has convincingly shown the importance of using the right resource subsidy to reduce crop damage. The results were obtained by using a predator-prey-pollen model (Rijn et al., 2002), and are likely to apply qualitatively to other pest-natural enemy systems.
When starting with a low initial predator density in the greenhouse (0.1 predators/plant), the lowest mean prey density was found with the highest pulse quantities and a long interval between pulses (Figure 5). The prey larvae threshold of 30 prey larvae/plant was not exceeded for this range. A negative effect of this pollen supply on pest densities occurred with a higher pulse quantity and long interval (apparent competition). The mean pest density was kept low due to an increase in predator density and because the prey density did not reach high levels, the predator density was reduced as well. The underlying dynamics are shown in Figures 2 & 3, where it can be seen that pollen supply had a short-term positive effect on the prey density, which increased after each pollen supply due to satiation of the predator. However, the negative effects of this pollen supply on the pest population were stronger than the positive effects, resulting in low average prey densities.
These positive effects of pollen supply on pest densities were stronger with an interval of one day and a high pulse quantity. Increased pollen supply caused satiation of the predators, resulting in a decrease of predation on the pest. This is not beneficial in preventing crop damage and shows the
importance of supplying the optimal quantity and frequency of pollen pulses, which, for an initial predator density of 0.1 predators/plant, is a long interval with higher pulse quantity.
Furthermore, the predator-prey-pollen model was tested with a higher initial predator density to determine if crop damage could be reduced even more than with a lower initial predator density. The mean total prey and prey larvae densities were lower with the release of higher densities of predators because a higher predator density consumed more prey and supplying pollen to this population resulted in less satiation (Figure 6). The prey larvae threshold was not exceeded for any pulse quantity or interval. Besides, a lower prey density is favourable in preventing crop damage and indicates a stronger negative effect of alternative food on the pest population.
In contrast to simulations with low initial predator densities, the lowest prey mean densities were now found with short intervals, and not with long intervals (cf. Figures 5 and 6), and a medium to high pulse quantities (Figure 6). A stronger negative effect of alternative food on the pest population occurred with increasing pulse quantity and shorter interval because food was abundant, which increased the predator densities and therefore the predation on prey. In contrast, a more positive effect occurred with a long interval and low pulse quantity. The prey and prey larvae density increased because the predator density decreased due to food scarcity.
To conclude, a higher initial predator density was more favourable in preventing crop damage than a lower initial predator density because the negative effects of alternative food on the pest population are stronger. Therefore, a higher initial predator density with medium to high pulse quantities and a long interval would be the most optimal because this alternative food supply results
in the lowest mean prey density, which is a measurement for crop damage.
From these results, it can be concluded that the highest prey and prey larvae densities occurred with a low initial predator density and with short intervals. In contrast, with a higher initial predator density, the highest prey and prey larvae densities occurred with long intervals. This shows that the optimal frequency and quantity of alternative food supply depends greatly on the initial density of the predators. Besides, there is an optimum in frequency and quantity of alternative food supply; more frequently or higher quantities are not always better and there is no clear strategy that works for all cases but it depends on factors such as the initial density.
4.2 Suggestions for future research
The demand for biological pest control in greenhouses has been increasing due to the negative impact of pesticides on the environment (Crall et al., 2018; van Lenteren et al., 2018). The findings of this research suggest that supplying alternative food to natural enemies is a good alternative in controlling pest densities because it results in apparent competition-like interactions between the alternative food and the pest (Holt & Bonsall, 2017; Rijn et al., 2002). However, before the obtained optimal frequency and quantity can be implemented in agriculture practices, it is of great importance that more research is done because the results show that alternative food supply in a predator-pest-pollen system often results in better pest control, but not always. Therefore, this model needs to be adapted to specific systems with differences in crop duration, guidelines of good damage prevention, and the predator’s and pest’s life cycles. In addition, the results of the current research need to be verified experimentally.
The predator-prey-pollen model was simulated with fixed initial prey and predator densities for each life stage. It was assumed that the pest species was present in the crop from day one, and only adult predators were introduced. However, another method of biological pest control is that the predator population is introduced with alternative food before a pest invasion. Thus, if the pest invades the predator could predate immediately, causing a decrease in crop damage (McMurtry & Scriven, 1966). To improve the optimal frequency and quantity of alternative food supply for this method of biological pest control, a simulation could be done where first the predator population reaches an average density based on the resource pulses before the pest population is added in the model. By determining the optimal predator density, the probability of successful invasions of pests could be maximally reduced.
In the predator-prey-pollen model used, the assumption was made that the generalist predator had no preference for the consumption of pollen or prey. It could be investigated what occurs to the system when the generalist predator prefers one of the food sources. If the predator favours pollen as a food source, the predator could prey less on the pest due to switching, causing the pest density to increase. In contrast, a preference for the pest would cause the predators to consume more pest than pollen, which would result in better pest control. The alternative food would then be important only in periods of low pest densities.
Furthermore, in this research, it is assumed that only the predator consumed alternative food. van Rijn et al. (2002) showed that when both the predatory mite and herbivorous thrips consumed alternative food, the thrips density decreased considerably when pollen grains were distributed locally on plants. Predators accumulated around pollen sites and predated on pest that came to
consume pollen, which decreased the benefits of pollen consumption for pest. Besides, Leman & Messelink (2015) state that it is important that the nutritional quality of food supply is more beneficial for the predator than for the pest when both predators and prey consume pollen grains. Therefore, the assumption, that pollen grains are also edible for pest, needs to be considered in the model for determining the optimal frequency and quantity of alternative food supply and perhaps also the way in which the alternative food is offered (concentrated or over the entire crop), as argued by van Rijn et al. (2002).
Lastly, a certain quantity and frequency of a food pulse could be the most beneficial for decreasing pest densities, however, economic aspects should be taken into account as well. Supplying alternative food with shorter intervals and at higher quantities is likely to be more expensive than a long interval and lower quantities due to higher material and labour costs. Therefore, a trade-off should be made between the lowest prey and prey larvae density and costs if a shorter interval and a higher pollen quantity are favourable in preventing crop damage.
5. Acknowledgements
I would like to thank Arne Janssen for supervising my research project and for all the feedback, Hanneke de Lange for working together on the model and the mental support during the pandemic, and Paul van Rijn for let me use and adjust his predator-prey-pollen model.
6. References
Bale, J. s, van Lenteren, J. c, & Bigler, F. (2008). Biological control and sustainable food production.
Philosophical Transactions of the Royal Society
B: Biological Sciences, 363(1492), 761–776.
https://doi.org/10.1098/rstb.2007.2182
Crall, J. D., Switzer, C. M., Oppenheimer, R. L., Versypt, A. N. F., Dey, B., Brown, A., Eyster, M., Guérin, C.,
Pierce, N. E., Combes, S. A., & Bivort, B. L. de. (2018). Neonicotinoid exposure disrupts bumblebee nest behavior, social networks, and thermoregulation. Science, 362(6415), 683–686. https://doi.org/10.1126/science.aat1598
Holt, R. D. (1977). Predation, apparent competition, and the structure of prey communities. Theoretical
Population Biology, 12(2), 197–229.
https://doi.org/10.1016/0040-5809(77)90042-9 Holt, R. D. (2008). Theoretical Perspectives on Resource Pulses. Ecology, 89(3), 671–681.
https://doi.org/10.1890/07-0348.1
Holt, R. D., & Bonsall, M. B. (2017). Apparent
Competition. 28.
Janssen, A., Pallini, A., Venzon, M., & Sabelis, M. W. (1998). Review Behaviour and indirect interactions in
food webs of plant-inhabiting arthropods.
Experimental & Applied Acarology, 22(9), 497–
521. https://doi.org/10.1023/A:1006089924336 Janssen, A., & Sabelis, M. W. (2015). Alternative food and biological control by generalist predatory mites: The
case of Amblyseius swirskii. Experimental and
Applied Acarology, 65(4), 413–418.
https://doi.org/10.1007/s10493-015-9901-8 Leman, A., & Messelink, G. J. (2015). Supplemental food that supports both predator and pest: A risk for biological control? Experimental and Applied
Acarology, 65(4), 511–524.
https://doi.org/10.1007/s10493-014-9859-y McMurtry, J. A., & Scriven, G. T. (1966). The Influence of Pollen and Prey Density on the Number of Prey
Phytoseiidae). Annals of the Entomological
Society of America, 59(1), 147–149.
https://doi.org/10.1093/aesa/59.1.147
Messelink, G. J., Bennison, J., Alomar, O., Ingegno, B. L., Tavella, L., Shipp, L., Palevsky, E., & Wäckers, F. L.
(2014). Approaches to conserving natural enemy populations in greenhouse crops: Current methods and future prospects.
BioControl, 59(4), 377–393.
https://doi.org/10.1007/s10526-014-9579-6 Messelink, G. J., Maanen, R. van, van Steenpaal, S. E. F., & Janssen, A. (2008). Biological control of thrips and
whiteflies by a shared predator: Two pests are better than one. Biological Control, 44(3), 372– 379.
https://doi.org/10.1016/j.biocontrol.2007.10.0 17
Nomikou, M., Sabelis, M. W., & Janssen, A. (2010). Pollen subsidies promote whitefly control through the
numerical response of predatory mites.
BioControl, 55(2), 253–260.
https://doi.org/10.1007/s10526-009-9233-x Rijn, P. C. J. van, Houten, Y. M. van, & Sabelis, M. W. (2002). How Plants Benefit from Providing Food to
Predators Even When It Is Also Edible to Herbivores. Ecology, 83(10), 2664–2679.
https://doi.org/10.1890/0012-9658(2002)083[2664:HPBFPF]2.0.CO;2 Sato, T., El‐Sabaawi, R. W., Campbell, K., Ohta, T., & Richardson, J. S. (2016). A test of the effects of timing of
a pulsed resource subsidy on stream ecosystems. Journal of Animal Ecology, 85(5), 1136–1146. https://doi.org/10.1111/1365-2656.12516
van Lenteren, J. C., Bolckmans, K., Köhl, J.,
Ravensberg, W. J., & Urbaneja, A. (2018). Biological control using
invertebrates and microorganisms: Plenty of new opportunities. BioControl, 63(1), 39–59. https://doi.org/10.1007/s10526-017-9801-4 Yamamuro, M., Komuro, T., Kamiya, H., Kato, T., Hasegawa, H., & Kameda, Y. (2019). Neonicotinoids disrupt aquatic food webs and decrease fishery yields. Science, 366(6465), 620– 623.
