Consistent Between-Individual Differences in

1 downloads 0 Views 453KB Size Report
Jun 7, 2018 - Tan and Tan 2017; Tan et al. 2017a). A reduced foraging perfor- mance owing to neural .... in height) filled to the brim with tap water. The total ...
Plant–Insect Interactions

Environmental Entomology, XX(X), 2018, 1–9 doi: 10.1093/ee/nvy087 Research

Consistent Between-Individual Differences in Foraging Performance in a Floriphilic Katydid in Response to Different Choices Ming Kai Tan,1 Fang Ni Goh, and Hugh Tiang Wah Tan Department of Biological Sciences, National University of Singapore, 14 Science Drive 4, Singapore 117543, Republic of Singapore and 1Corresponding author, e-mail: [email protected] Subject Editor: Joe Louis Received 11 March 2018; Editorial decision 14 May 2018

Abstract The neural constraint hypothesis is one of the central ideas for the understanding of insect–plant interaction but there are still knowledge gaps in the data for foraging behavior and the performance of herbivores, and particularly florivores. We used a floriphilic katydid, Phaneroptera brevis (Serville, 1838) (Orthoptera:Tettigoniidae) and a naturalized weed, Bidens pilosa L. (Asteraceae) in caged experiments in an insectary to answer these questions: 1) How does the foraging performance of the floriphilic katydid vary when exposed to a choice in the number of capitula and types of florets of B. pilosa? 2) Does the foraging performance of the katydid, when exposed to multiple choices, improve with time, and are between-individual differences in foraging performance consistent? We observed that having more choices in the floret types and number of capitula is generally associated with a reduced foraging performance of the katydids. Floret types and number of capitula, however, did not have an interactive effect on foraging performance. We also found that the differences in foraging performance in response to choice tend to be consistent between katydids but each katydid became more efficient and decisive over time.That learning and experience can improve the foraging performance of the katydid has provided us with some insights as to how a continuum of efficient and inefficient katydids can be maintained in a population. Key words: behavioural ecology, insect–plant interaction, florivory, learning, neural constraint hypothesis

Florivory—the feeding of flowers—remains less well studied in ecology (McCall et al. 2018). Despite the fact that flowers are essential for sexual reproduction and reproductive fitness in plants, and that florivory can directly reduce reproductive success, and/or quantity and quality of pollinator visits (McCall and Irwin 2006) by disrupting flowering times (Kawagoe and Kudoh 2010) and reducing attractiveness to pollinators (Strauss 1997, Lehtila and Strauss 1999), studies on flower–insect interaction tend to be focused on pollination. To ensure a more holistic understanding of flower– insect interaction, it is crucial to study how florivores behave and respond to floral resources, since pollinators and florivores can utilize different cues from different floral parts to assess the quality of the flower (Irwin et  al. 2004, Gegear et  al. 2017). Floral morphology is usually more complex than leaf morphology. However, a single flower is usually treated as a single resource, even when it is unlikely that florivores feed on all the different floral parts indiscriminately. Florivores can target specific structures (e.g., pedicel, petal, sepal, stamen, gynoecium) for different nutrients (Junker and Parachnowitsch 2015). To our best understanding, there has been little work done on the behavior of florivores, including how

a florivore responds to different resource types within a flower (Strauss and Whittall 2006). Foraging performance of a florivore, especially generalists, can be reduced by the presence of different resource types within a flower. The neural constraint hypothesis predicts that generalist, phytophagous insects have a more limited neural capacity and poorer foraging performance than their specialist counterparts in response to choice based on different available resources (Bernays and Wcislo 1994, Bernays and Funk 1999, Nylin et al. 2000, Bernays 2001). A foraging generalist insect can also be distracted and confused (through reduced attention to the task at hand, so leading to a lower efficiency and accuracy in discriminating differences in resource quality) when faced with more choices (Bernays 1999, Bernays and Bright 2001, Tan and Tan 2017; Tan et  al. 2017a). A  reduced foraging performance owing to neural constraint can be indirectly undesirable to the survival and reproductive success of individuals. For example, being inefficient and inattentive during foraging can expose the individual to greater predation risk (Bernays 2001). Likewise, being less accurate in decision-making can also lead to feeding on less nutritious or more toxic food (Bernays 2001). This has even led to the

© The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: [email protected].

Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy087/5033848 by NATIONAL UNIVERSITY OF SINGAPORE MEDICAL LIBRARY user on 07 June 2018

1

Environmental Entomology, 2018, Vol. XX, No. X

2 suggestion that the neural constraint that generalists face is a possible underlying reason why there is a greater abundance of specialists in forests rich in host plant species (Bernays and Wcislo 1994, Bernays 1996, Bernays 2001, Agrawal et al. 2006). While some research on the neural constraint hypothesis and florivory has been done (see Tan and Tan 2017, Tan et  al. 2017a, Tan et  al. 2018), there is still insufficient information on how the foraging performance of florivores is associated with resource choices, when faced with multiple choices. It was demonstrated for the first time that a polyphagous katydid forages inefficiently and inattentively when exposed to a high density of flowers (see Tan et al. 2017a). Additionally, the same katydid species is also less efficient and accurate in foraging when exposed to flowers and leaves than when exposed to only flowers (see Tan and Tan 2017). Nonetheless, investigating how different resource types and density within a single host plant part (e.g., flower) can influence behaviors of flower-visitors has not been done. Moreover, in addition to resource availability, behavior and foraging performance of florivores may be individual specific in a population. It is unclear, when faced with multiple choices, whether the individuals in a population with a lower foraging performance are always consistently poorer than counterparts from the same population with a higher foraging efficiency. Since foraging performance involves decision-making and learning (i.e., cognition) and personality is the consistent inter-individual differences in behavior across time and/or context, investigating how consistent the differences are in the foraging performance between individuals in a population can help understand how animal personality types and cognition are linked (Sih and Del Giudice 2012, Guenther and Brust 2017, Tan et  al. 2018). Additionally, how individuals in a population can improve in their foraging performance over time remains understudied. An individual’s improvement in foraging performance over time can be attributed to learning and experience, but studies on this are also limited to a few flower-visiting insect species (Bernays 2001, Guenther and Brust 2017). To investigate these research questions, one can use repeatability estimates to investigate between-individual variations in behavior, or more specifically foraging performance, of animals (Nakagawa and Schielzeth 2010). By also examining how the foraging performance varies across the repeated trials, we can examine if the katydids’ foraging performance improves. By studying the behavior of florivores, which have been understudied in general, we aim to contribute to a more holistic understanding of florivore–flower interactions from

both the perspectives of the plant and insect (McCall and Irwin 2006). Specifically, to address deficiencies in the knowledge on how a florivore responds to different resource types in a flower, we performed laboratory-based experiments using a polyphagous, floriphilic, Southeast Asian katydid, Phaneroptera brevis (Serville, 1838), and the North to Central American Bidens pilosa L.  (Asteraceae; Ballard 1986), a naturalized weed of wasteland vegetation in Singapore (Fig.  1). We examined four proxies of foraging performance, namely efficiency in feeding, restlessness in foraging, indecisiveness in foraging, and acceptability of food. By repeatedly subjecting each katydid individual to the same treatment over a few days (different treatments consisted of different number of capitula and types of florets in a capitulum of B. pilosa), we aimed to answer the following research questions: 1) How does the foraging performance of the floriphilic katydid vary when exposed to choice in the number of capitula and types of florets of B. pilosa? We predicted that the katydid will consider the ray and disc florets as different resource types and that high densities of each floret type can lead it to display reduced efficiency, attentiveness, and acceptability of food, as well as more indecisiveness. 2)  Does the foraging performance of the katydid, when exposed to multiple choices, improve with time, and are between-individual differences in foraging performance consistent? We predicted that efficiency and decision-making will improve with time (Bernays 2001) but a less decisive or efficient katydid individual will remain less decisive and efficient than counterparts from the same population (i.e., consistent between-individual differences in foraging performance).

Material and Methods Study Subjects Phaneroptera brevis is a common flower-visitor and feeds on the flowers of  numerous species of flowering plants, including those of B.  pilosa (Tan et  al. 2017b) which is an important species, as although it can be an invasive weed in some parts of the world, it can also be an important food source for many local pollinators (Lowe et  al. 2000; Tan et  al. 2017b). Each capitulum of B.  pilosa, as is typical of those of many Asteracaeae members, consists of a central disc (containing numerous, yellow disc florets) and white ray florets around the disc (Ballard 1986) (Fig. 2). While the soft (less tough) ray florets provide an easy-to-eat food for the katydid, the disc florets contain the pollen grains that provide the proteins that are essential for the katydid.

Fig. 1.  Phaneroptera brevis feeding on a Bidens pilosa capitulum with ray florets (a) and without (b), under natural conditions.

Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy087/5033848 by NATIONAL UNIVERSITY OF SINGAPORE MEDICAL LIBRARY user on 07 June 2018

Environmental Entomology, 2018, Vol. XX, No. X

3

Fig. 2.  Bidens pilosa capitulum and the two types of florets: ray and disc florets. Scale bar = 10 mm.

Table 1.  Summary of different treatments that the individuals of the katydid, Phaneroptera brevis were subjected to Treatment

1 2 3 4 5 6

Replicates

Gender

Number of capitula

Total = 60

Male

Female

10 10 10 10 10 10

5 5 5 5 5 5

5 5 5 5 5 5

Behavioral Assay Between January and June 2017, experiments were carried out at the National University of Singapore Department of Biological Sciences Insectary. We opportunistically collected adult P.  brevis individuals and stem-cuttings of B.  pilosa from a wasteland site of about 6,500 m2 along Lorong Lada Hitam, off Mandai Road, Singapore (N1.41846°, E103.79164°). This was done in the late afternoon (4–5 pm) when it was cooler to minimize water stress for the plants. Upon returning to the insectary, we placed each katydid inside a standardized, free-standing, insect cage (25  cm in diameter and 33  cm in height) for approximately 19  h for acclimatization and fasting to clear out the gut content (Tan et al. 2017a). The amount of time from collection of katydid the to the start of experiment the following day was approximately nineteen hours. The study subjects were subjected to a 9:15 light–dark cycle. Only moistened tissue paper was provided to prevent dehydration. After the acclimatization and fasting period, we randomly assigned the katydids to different treatments consisting of varying numbers of capitula (density) and different types of available florets (Table 1). Sixty adult katydids were collected and used for the experiments, 30 of which were males and 30 of which were females. We placed either two, four, or six stem-cuttings of B.  pilosa in a standardized plastic container (4.5  cm in diameter and 8.5  cm in height) filled to the brim with tap water. The total number of capitula was provided ad lib such that there were always more than enough for a katydid to eat, and as confirmation, none of the katydids was observed to have eaten all capitula within the experiment’s duration. Each stem-cutting bore only one capitulum. The capitula were arranged haphazardly within the insect cage, where

Choice of floret types Ray florets

2 2 4 4 6 6

Present Removed Present Removed Present Removed

Disc florets Present Present Present Present Present Present

the katydids were free to move and forage between and within the capitula (Fig.  3). The leaves, immature capitula, and capitula with developing or developed achenes were all removed before use of the experiment. The height of the capitula above the water level was standardized to be 12 cm as the height of the capitula may influence the foraging behavior of the katydids. For treatments where the ray florets were absent, we carefully removed all ray florets from all the capitula. We did not use treatments for which only ray florets were present because such a capitulum usually does not occur in nature, and it was not possible to remove the disc florets without damaging the ray florets. The experiment commenced upon introducing the plant material into the cage. Each experiment was video-recorded using a full high-definition video recorder (Panasonic HC-V180 camcorder) for 5 h. To quantify foraging performance of each katydid, we analyzed the video recordings frame by frame to observe the behavior of the katydids: 1) The number of feeding bouts and meals. A feeding bout is defined as a period of continuous feeding on the same capitulum whereas a meal is defined as a series of feeding bouts on the same capitulum (Bernays and Bright 2001). 2) Durations of meals and feeding bouts on each floret type. 3)  Time spent pausing between feeding bouts within a meal. At the end of the 5 h, we removed the plant material and the katydids were left to fast so as to clear the gut content for another 19 h (the same as the first day prior to the start of the experiment). Thereafter, new plant material was provided again and the experiment video-recorded for another 5 h. To investigate how the foraging performance of the katydid, when exposed to multiple choices, change with repeated exposure and how the differences in foraging performance were consistent between individuals, we repeated the

Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy087/5033848 by NATIONAL UNIVERSITY OF SINGAPORE MEDICAL LIBRARY user on 07 June 2018

Environmental Entomology, 2018, Vol. XX, No. X

4

Fig. 3.  Experimental set up showing the insect cage and the stem-cuttings of Bidens pilosa distributed within it. One katydid was placed inside the cage to observe its foraging behavior.

above trials over 3 d. Each individual katydid was subjected to the same treatment throughout the three trials.

Table 2.  Pearson’s correlation coefficients between the three proxies: efficiency, restlessness, and indecisiveness of the katydid Efficiency

Data Analyses We first compared the katydids’ preference for ray and disc florets using the paired Wilcoxon signed rank test with continuity correction. To investigate how the foraging performance of the katydids varied with the interactive effect of the number of capitula and different floret types, we used these four measures as proxies for foraging performance: 1)  Efficiency was quantified as the average feeding time per capitulum. The katydid can be less efficient and spend less time feeding on each resource type in the presence of more resources (i.e., more choices) (Bernays 1999, Bernays and Bright 2001). 2) Restlessness within each meal was quantified as the average number of bouts per meal. In the presence of more choices, the katydid can be more restless (or less attentive) with an interrupted feeding pattern within each meal, leading to a shorter bout length and hence more separate bouts within a single meal (Bernays 1999, Bernays and Bright 2001). 3) Indecisiveness was quantified as pause time, that is, the proportion of nonfeeding time to the total time nonfeeding and feeding. In the presence of more choices, the katydid can have difficulty making a decision and spend more time moving around foraging for alternative food types and pausing instead of continuously feeding (Bernays and Bright 2001). 4)  Acceptability of suitable food was quantified as the probability of whether the katydid accepted (i.e., consumed) the food during the 5-h trial. In the presence of more choices, the katydid can reject (i.e., do not consume) food that is typically acceptable owing to difficulty and mistakes in decision-making (Bernays et al. 2004). We did not find strong correlations between the four proxies for foraging performance (Table 2). To investigate how foraging performance of the floriphilic katydids varied when exposed to a varying number of choices owing to different number of capitula and types of florets in a capitulum of B. pilosa, we proposed and compared four biologically meaningful

Efficiency Restlessness

Restlessness

Indecisiveness

0.14

−0.38 0.35

Correlations between the three proxies and acceptability were not included because efficiency, restlessness, and indecisiveness are not applicable if the katydids did not accept the food.

models: 1)  absence/presence of two types of florets, 2)  number of capitula, 3)  absence/presence of two types of florets + number of capitula, and 4) absence/presence of two types of florets × number of capitula or ray florets. A null model (1) was also compared. To investigate how the foraging performance of the katydid, when exposed to multiple choices, changes with time and how the differences in foraging performance were consistent between individuals, trial number was also fitted as a fixed effect and individual katydid identity was fitted as a random intercept to examine within- and between-individual variations in behavior, respectively. The adjusted repeatability (i.e., repeatability after controlling for fixed effect) was calculated on the original scale as between-individual variance divided by the sum of between-individual variance and residual variance (Nakagawa and Schielzeth 2010). The intraclass  correlation coefficient (ICC), an estimate of repeatability, was reported. We also performed 500 parametric bootstraps to obtain the 95% confidence intervals for the random effect. This was done using the ‘rpt’, ‘rptProportion’, and ‘rptBinary’ functions in the R package rptR (Schielzeth and Nakagawa 2011). Each measure of foraging performance was fitted separately. Efficiency and restlessness (both continuous variables) were fitted with linear mixed-effects models using the ‘lmer’ function, and indecisiveness and acceptability (both proportion variables) were fitted

Downloaded from https://academic.oup.com/ee/advance-article-abstract/doi/10.1093/ee/nvy087/5033848 by NATIONAL UNIVERSITY OF SINGAPORE MEDICAL LIBRARY user on 07 June 2018

Environmental Entomology, 2018, Vol. XX, No. X with generalized linear mixed-effects models (GLMMs) with the binomial error via the log-link function using the ‘glmer function from the R package lme4 (Bates et  al. 2014). For each measure, models were ranked accordingly to the small-sample-size corrected version of Akaike information criterion (AICc) using the ‘MuMIn’ package (Barton and Barton 2015). Models with the difference between the AICc of a particular model and that of the best model less than 2.0 (Burnham and Anderson 2002) were interpreted together. We obtained the confidence interval and estimates for each predictor in these models. Overdispersion was checked for by comparing the residual deviance to the degree of freedom. Heteroscedasticity was also checked. The marginal and conditional R2 values (i.e., R2m and R2c, respectively) were obtained using the ‘r.squaredGLMM’ function from using the R package ‘MuMIn’. All statistical analyses were conducted using R software v.3.4.3 (R Core Team 2018).

Ethical Note We were careful not to harm the katydids during the collection, housing, and experiments. We were careful to minimize the impact on their natural habitat during their collection. We released the katydids back into their original habitat after the experiment.

Results We found that the katydids fed on both the disc and ray florets. However, the katydids generally preferred the disc florets over the ray florets and spent significantly more time feeding on the disc

5 florets (difference between median feeding time on the disc and ray florets = 9.5, V = 38, P-value < 0.001, n = 90). This indicated that the katydids did consider the disc and ray florets as two different food resources. For the effect of varying the number of choices on the efficiency of the katydid, the best model contained multiplicative interaction between the number of capitula and the absence/presence of two types of florets (R2m = 0.11, R2c = 0.34) (Table 3). The interaction effect was not strong (estimate  =  2.25; 95% CI −1.86, 6.35) but increasing number of capitula had a strong negative effect on the efficiency of the katydid (estimate  =  −4.25; 95% CI −7.15, −1.35) (Fig. 4; Table 4). For the effect of varying number of choices on the restlessness of the katydid, the two best models (with delta