Specialization and accuracy of host-searching butterflies in complex and simple environments

Specialization and accuracy of host-searching butterflies in complex and simple environments Abstract Populations that have access to a variety of resources are often composed of individuals that specialize on different subsets of resources. Understanding the behavioral mechanisms that drive such individual specialization will help us predict the strength of this specialization across different environments. Here, we explore the idea that individual specialization may be a consequence of constraints on an individual’s ability to process information. Because many environments contain an overwhelming number of resources and associated stimuli, individuals that specialize by focusing on only a subset of these resources may make more accurate decisions than individuals that generalize. Furthermore, we expect individuals in complex environments, where there are more resources and associated stimuli to process, to specialize during their search for resources compared with individuals in simple environments. We tested these predictions by measuring the accuracy and degree of specialization of naïve cabbage white butterflies (Pieris rapae) searching for 2 target host species (radish and cabbage) in simple (1 distractor species) and more complex (4 distractor species) environments. Only butterflies that specialized on cabbage were more accurate than butterflies that visited a mixture of both radish and cabbage. Furthermore, naïve butterflies searching for hosts in complex environments did not adopt more specialized foraging strategies than naive butterflies searching for hosts in simple environments. Taken together, these results suggest that the foraging benefits associated with specialization might only apply to certain resources (perhaps those that have readily recognizable cues) and that such specializations can be related to accuracy across multiple environments. INTRODUCTION Individuals in a population often use only a small subset of the resources available to the entire population (Van Valen 1965; Fox and Morrow 1981; Bolnick et al. 2003). This pattern of resource use, termed individual specialization, has been observed across a variety of taxa and can have ecological and evolutionary consequences (Estes et al. 2003; Dall et al. 2012; Bolnick et al. 2011). The degree to which individuals in a population differ in their resource use can affect population dynamics and stability (Lomnicki 1978; Hughes et al. 2008), species interaction networks (e.g. pollinators in Brosi 2016), intra- and inter-specific competition (Svanbäck and Bonlick 2007; Duffy 2010), and may ultimately result in sympatric speciation (Dieckmann and Doebeli 1999). Individual specialization also has consequences for the individual organism. Specializing on only one resource can affect an individual’s foraging performance (Bernays 1998), vulnerability to predation (Darimont et al. 2007), and risk of parasite exposure (Reimchen and Nosil 2001; Johnson et al. 2009). Over the past decade, research on individual specialization has moved from documenting the presence or absence of individual specialization in nature to exploring specific hypotheses regarding variation in specialization over time and space (Araújo et al. 2011; Dall et al. 2012; Layman et al. 2015). To make more explicit predictions about how and why specialization changes across ecological gradients, we need to better understand the behavioral mechanisms that promote and facilitate individual resource specializations. From a behavioral perspective, specialization may arise from limitations on the amount of information an organism can process (Bernays 2001; Dukas 2002). The environment contains more relevant information than any one individual can actually evaluate and, consequently, many animals (including humans) must focus on specific incoming signals and ignore others in order to make choices efficiently (Dusenbery 1992; Kastner and Ungerleider 2000). When humans, monkeys, and birds are faced with the task of detecting multiple stimuli, focusing attention on one of these stimuli increases the probability of detecting that stimulus (Moran and Desimone 1985; Kastner and Ungerleider 2000; Dukas and Kamil 2001). This idea, that there are costs to divided attention, is often referred to as “neural constraints,” “limited attention,” or the “information-processing hypothesis” (Bernays 2001; Dukas 2004; Egan and Funk 2006) and has implications for how animals make resource-related choices. Because focusing on a small number of resources requires less attentional load than focusing on a large number resources, the information-processing hypothesis predicts that specialists make more efficient decisions than generalists (Bernays and Wcislo 1994; Dall and Cuthill 1997; Bernays 2001). Specializing on a subset of available resources could increase a forager’s speed and/or accuracy (ability to distinguish between rewarding and nonrewarding resources), 2 key components of efficient decision-making (Chittka et al. 2009). Field studies have lent support to the idea that specialists are more accurate than generalists by testing this prediction across species, showing that specialist species are often more accurate and efficient than generalist species (Janz and Nylin 1997; Bernays, Singer, and Rodrigues 2004; Terraube et al. 2011). However, variation in specialization may also occur among individuals within a population (Fox and Morrow 1981; Bolnick et al. 2003). For example, certain individual bumblebee workers foraging in the same area might visit only one species of flower while other individuals might visit three species of flowers (Heinrich 1976). This example of resource specialization is a flexible behavioral strategy that can vary within and among individuals as well as across ecological contexts (Brosi 2016). Constraints on information processing could contribute to this individual specialization; however, there has been less exploration as to whether specialized individuals in a population receive the same decision-making advantages we see in specialized species. The studies that do address selective information processing on the individual level present mixed support for the idea that more specialized individuals make more accurate and efficient decisions than more generalized individuals. Egan and Funk (2006) found that specialized leaf beetle “host forms” were more accurate than generalist forms and Janz (2003) found that polyphagous members of the butterfly Polygonum c-album had longer decision times than specialists. However, in the butterfly Euphydryas editha, natural variation in specialization among individuals was not associated with faster host recognition or better accuracy (Wee and Singer 2007). Given these conflicting results, our work will address whether and under what conditions selective processing of information contributes to resource specialization on the individual level. The information-processing hypothesis not only predicts consequences for specialized and generalized search strategies, it also has implications for how individuals interact with environments that vary in cognitive complexity. A complex environment can make resource decisions more challenging by decreasing the ratio of resources (called “targets”) to nonresources (called “distractors”), increasing the similarity between targets and distractors (“discriminability”) and increasing the similarity between targets/distractors and the background in which targets and distractors are embedded (“detectability”). Each of these different types of complexity can reduce the accuracy and efficiency of individuals that are searching for resources (Perfecto and Vet 2003; Gols et al. 2005; Xiao and Cuthill 2016). If resource-related decisions become more difficult in complex environments, specialists, because they have the opportunity to simplify their decisions, may have an advantage. Thus, we can make two predictions about how information processing might influence individual search behavior in complex environments. Firstly, individuals searching in complex environments might be expected to adopt a more specialized foraging strategy than different individuals searching in simple environments. Secondly, more specialized individuals are expected to make more accurate decisions than more generalized individuals. To explore these predictions, we observed individual naïve cabbage white butterflies (Pieris rapae) searching for 2 different species of host plants embedded in either complex (multiple species of nonhost plants) or simple (one species of nonhost plant) communities. In the field, cabbage white butterflies lay eggs on a variety of host plant species in the family Brassicaceae (Scott 1986). In the lab, we further observed that individual butterflies vary along a continuum where some individuals oviposit on multiple host species and others oviposit primarily on one host species, even when viable alternate hosts are available. Because cabbage whites are relatively small and display fairly typical host searching behavior in large greenhouse flight cages, we can observe the behavior of a single butterfly and calculate both the individual’s degree of host specialization when offered multiple hosts and the individual’s foraging accuracy. Although insect specialization can be measured across a variety of life stages, contexts, and timescales, we measure specialization in the present study as the tendency of an adult female butterfly to lay her eggs on primarily one host plant species (more specialized) rather than multiple host plant species (more generalized) during one bout of host searching. If limits to information processing promote individual specialization in this system, we first predict that more specialized butterflies (those that visit only one of the 2 host species) will be more accurate than more generalized butterflies (those that visit both host species). If complex environments place further strain on information processing, we expect the individual butterflies searching for hosts in complex environments (with multiple species of nonhost plants) to be more specialized than the individual butterflies searching for hosts in simple environments (with one species of nonhost plant). METHODS Care for experimental butterflies Adult Pieris rapae used in this experiment were the offspring of over forty wild-caught females captured in Saint Paul, Minnesota during Summer 2014. Larvae were reared on artificial diet (recipe in Jaumann and Snell-Rood 2017) in a 14:10 h light:day climate chamber at 23 °C and approximately 60% relative humidity. When butterflies emerged from their chrysalides, they were immediately marked with an identification number on their hind wing and stored in glassine envelopes in a refrigerator at 5 °Celsius for no more than four days. Butterflies were subsequently transferred to 60 cm3 “bug dorm” mesh cages where they were allowed to mate with males for 48 h on an ad lib diet of 10% honey water solution before behavioral testing in our experimental plant arrays. Plant care and measurements Cabbage (Brassica oleraceae var. Earliana) and radish (Raphanus sativus var. Cherry Belle) were selected as host plants because both Brassica and Raphanus species are commonly planted on farms, can be found in organic gardens, and have related feral species commonly used by the cabbage white butterfly (Scott 1986). Radish and cabbage have relatively similar growing times and similar leaf area at 40 days of growth. Basil, Swiss chard, lettuce and pepper (basil: Ocimum basilicum, Swiss chard: Beta vulgaris var. Fordhook®Giant, lettuce: Lactuca sativa var. Nevada, pepper Capsicum annuum var. California Wonder) were selected as nonhost plants because they were similar in size and growing time to hosts but added a wide variety of leaf colors, shapes, and smells to environment. These nonhost plants are also commonly found in agricultural settings and would likely be present in the human-dominated disturbed landscapes where cabbage whites are most abundant (Kocher and Williams 2000). All flowers and buds on plants were removed and tall nonhost plants were cut to match host plant size to standardize structural complexity across trials. Seeds were first sown in May 2014, and new seeds were germinated every three weeks through late July. Seeds were germinated in pit trays and transplanted after 2 weeks to (5” × 5”) 2.65-liter pots filled with Sungro® potting mix (70–80% Sphagnum). One week after transplanting, all plants were fertilized with 6 g of 14-14-14 Osmocote® fertilizer and treated with OHP Marathon 1% Granular (a systemic pesticide with 1% Imidacloprid) to control greenhouse pest damage. Plants were kept in a greenhouse and watered daily. To compare color differences among our focal plants, 4 plants of each species were haphazardly selected for spectral analysis with an Ocean Optics Inc. JAZ-PX spectrometer. We calculated the mean reflectance and standard error of each species from 300 to 700 nm, the spectral sensitivity range of the cabbage white butterfly (Shimohigashi and Tominaga 1991). Reflectance was graphed with the “pavo” package in R, using a smoothing spectra with a span of 0.2 (Maia et al. 2013). Experimental arrays Adult cabbage white butterflies were allowed to search for host plants in one of 3 treatment arrays that varied in community complexity (one complex environment with 4 nonhosts and 2 simple environments, each with a single different nonhost species). Each treatment array contained a total of 24 potted plants, 12 of which were host plants (6 cabbage, 6 radish) and 12 of which were nonhost plants. These potted plants were arranged in a 6 × 4 rectangular array approximately 12 cm apart with no overlapping foliage inside a 1.6 × 2m flight cage. The number and identity of host plants (6 radish, 6 cabbage) remained constant over all 3 treatment arrays. Potted host plants were placed such that all immediate neighboring pots were of a different species. Adjacent neighbors were always nonhosts and diagonal neighbors were hosts of the other species. This design was chosen in an effort to minimize the patchiness inherent in the haphazard placement of plants. To manipulate environmental complexity, we changed the identity of the 12 nonhosts in each of the three treatment arrays. The “complex” treatment contained 4 different nonhost species (3 plants each of basil, Swiss chard, lettuce, and pepper) whereas the 2 “simple” treatments contained only one nonhost species (12 basil plants and 12 Swiss chard plants respectively). Nonhosts were placed such that members of the same species were as far apart from each other as possible. We chose to manipulate complexity by increasing nonhost diversity because such a manipulation increases the overall diversity of plants (and associated stimuli) in the environment without changing the host to nonhost ratio, the relative frequency of host plants, or the identity of host plants. Behavioral observations Butterflies were randomly assigned to a complexity treatment and tested one at a time by placing a focal individual on a randomly assigned host plant, in order to motivate host searching. Butterflies that did not immediately begin searching were not tested. The butterflies that immediately began searching were observed until they completed 20 host landings (landings on either cabbage or radish). Butterfly “landings” were characterized by tarsal drumming on the leaf surface, a behavior associated with chemoreception using tarsal receptors (Renwick and Chew 1994). Other behaviors during which butterflies were in contact with the plants (such as basking or resting) could be distinguished from “landings” because they are not accompanied by drumming and are marked by prolonged still behavior with wings held closed or slightly open at a 45 degree angle. After a host plant landing, females often lay a single egg and then continue to search for plants. Landings on both cabbage and radish result in an egg being laid 87.1% and 87.3% of the time respectively (for Ncabbage landings = 582 and Nradish landings = 458, 2015 data). One observer (MS) recorded landings on host and nonhost plants, the species of the plant, and the occurrence of drumming and oviposition. Re-visits to plants were counted as a new landing if butterflies flew over a neighboring plant in between. Each behavior and the time it occurred during the trial were recorded on JWatcher software installed on a laptop computer. No more than four butterflies were tested sequentially per treatment and the order of the treatments was changed on a daily basis. The eggs laid by each butterfly were gently brushed off the host plants after every trial to minimize the searching butterfly’s use of conspecific cues (Raitanen et al. 2014). Out of 96 butterflies that were initially released into the assay, a total of 37 butterflies successfully completed 20 host landings and were retained for analysis. Observations took place in a flight cage (2 × 1.6 × 1.6 m high PVC frame covered with nontreated mosquito netting) in the plant growth facilities greenhouses at the University of Minnesota between the hours of 10 am and 3 pm on sunny days from 30 June 2014 to 15 August 2014. Behavioral analyses Each female’s behavior was analyzed over her 20 host landings to calculate total host specialization, host landing preference, and host-finding accuracy. Host specialization and host landing preference were both calculated from a butterfly’s landings on host plants and served to situate each individual along a continuous gradient of resource use rather than as a binary “generalist” or “specialist.” However, host specialization and host landing preference each measure a slightly different aspect of host use. “Host specialization” measures a butterfly’s tendency to land primarily on one host species (regardless of whether this species is cabbage or radish) compared to two host species, spanning a gradient where more specialized individuals (landing on one species) represent one extreme and more generalized individuals (landing on 2 species) represent the other. Host specialization was estimated by calculating how far a butterfly’s landings on cabbage and radish departs from the expected number of landings given the number of cabbages in the assay (represented as 0.5 in the following equation).  host specialization = | (landings on cabbage) / (20 landings) − 0.5 | This metric ranges from 0 to 0.5, where 0.5 indicates that a butterfly landed either 20 times on cabbage or 20 times on radish, and 0 indicates that a butterfly landed on 10 cabbages and 10 radishes. There are many indices recommended for quantifying individual specialization that focus on how specialized an individual’s resource use is compared to the resource use of the population (outlined in Bolnick et al. 2002). We choose to follow the formula above because we want to focus on a measure of specialization that calculates how far an individual departs from resource availability in the controlled experimental environment. In contrast, “host landing preference” measures a butterfly’s tendency to land on cabbage compared to radish, a gradient where individuals that land primarily on cabbage are on one extreme and individuals that land primarily on radish are on the other, with the more generalized individuals falling in the middle of this axis. Host landing preference was estimated by counting the total number of cabbage landings out of 20 host landings, leading to a score from 0 to 20 where 0 indicates that all 20 host landings were made on radish and 20 indicates that all host landings were made on cabbage. We quantified a butterfly’s host-finding accuracy by counting the number “inaccurate” landings a butterfly made, i.e. the number nonhost plants a female drummed during the course of a foraging trial. Since all butterflies were allowed to make the same number of host landings and often laid eggs immediately following these landings, we assume that each butterfly had the opportunity to lay a similar number of eggs and that variation in search performance among butterflies is due largely to nonhost landings. To the extent that nonhost landings increase the time spent host searching (at the expense of other activities), and decrease the number of eggs laid during an oviposition bout, butterflies that land on more nonhosts would be expected to lay fewer eggs over their lifetime (Snell-Rood and Papaj 2009). To approximate the each butterfly’s total search time, we isolated the portions of a trial during which a butterfly was actively searching for or interacting with host plants (i.e. only drumming and oviposition behaviors from our ethogram) and added these portions together. Larval performance To estimate the consequences of a female’s oviposition choice, we raised cabbage white larvae on an ad lib diet of entire radish plants or cabbage plants and calculated their growth rate (pupal mass/development time). Gaining mass and developing quickly are considered advantages for butterflies because body size is often positively correlated with fecundity across species (García-Barros 2000; but see Bauerfiend and Fischer 2008 for a discussion of variation within species) and short development times allow larvae to minimize their vulnerability to predators while feeding (Heinrich 1993, Bernays 1997). To collect eggs for rearing, we allowed 10 wild-caught butterflies to oviposit on cabbage in one cage and 10 different butterflies to oviposit on radish in another cage, for 6 h. The plants with eggs were kept in these cages until they hatched and the larvae were large enough to transfer without injury (about 7 days after laying). Larvae were then distributed to new cages with fresh cabbage plants and fresh radish plants respectively so that there were 20 larvae per cage. The plants used for larval rearing were grown according to the same protocol as plants for behavioral trials except that no insecticide was applied. New plants were supplied to all rearing cages as needed over the course of larval development to provide an ad lib diet. The cages were monitored daily and once caterpillars formed pupae, they were collected and weighed on a microbalance. To calculate development time, we counted the number of days from the date the egg was laid to the date the larvae was collected. Growth rate was calculated for each larva by dividing its pupal weight by its development time. We repeated this experiment in four rounds throughout the month of August 2015. Statistics We used generalized linear regression models (GLM) for all of our analyses using R version 3.0.2 (R Core Team 2013). To test whether specialist butterflies were more accurate than generalist butterflies, we modeled how accuracy (nonhost landings) was affected by either specialization or landing preference (respectively) and nonhost community complexity. Specialization and landing preference were never used in the same model because they are both metrics calculated from the same raw data: a butterfly’s landings on cabbage versus radish. Accuracy was modeled using a Poisson distribution because nonhost landings were counted within a defined behavioral set (20 host landings) and had no upper bound. Because there could feasibly be interactions between nonhost complexity and specialization/landing preference on nonhost landings, we began with the full model and used corrected Akaike information criteria (AICc) and Akaike weights to compare the full model to simpler models (“AICcmodavg” package in R; Mazerolle 2015). We ran Tukey’s pairwise contrasts on the best-ranked model to identify differences between complexity treatments (“multcomp” package in R; Hothorn, Bretz, and Westfall 2008). To test whether P. rapae changed its oviposition behavior with increasing environmental complexity, we first evaluated how specialization was affected by nonhost complexity treatment using a GLM with Gaussian error structure. We also evaluated how landing preference was affected by nonhost complexity. When using host landing preference as a response variable (rather than as dependent variable as in earlier models), we calculated landing preference as the probability of landing on x number of cabbages in a sequence of 20 host landings. These models used binomial error structure with cabbage landings arbitrarily scored as “successes.” We were also interested in testing the consistency of landing preference in individual butterflies over time. To understand how landing preference changed over the course of a foraging bout, we divided a butterfly’s host landings into 2 periods: landings 1–10 and 11–20. We then modeled landing preference probability (probability of landing on x cabbages in 20 host landings) as a function of period and complexity treatment with individual as a random effect (“lme4” package in R; Bates et al. 2014). We also evaluated landing preference for cabbage or radish as a function of continuous landings within a trial using logistic regression, as follows. Each landing on a host (radish or cabbage) was assigned a number 1–20 indicating its position in the order of landings. For each landing, we scored cabbage as 1 and radish as 0 and modeled the probability of landing on cabbage as a function of landing number and plant community complexity with individual as a random effect. Total search time was modeled as a function of a butterfly’s accuracy, plant community complexity, and their interaction using a GLM with a Gamma distribution and log link. We evaluated the significance of terms with type-III likelihood-ratio tests (“car” package; Fox and Weisberg 2011). To test whether butterflies grew at a faster rate on cabbage or on radish, we modeled larval growth rate as a function of host type. RESULTS A total of 37 butterflies successfully completed foraging trials across three environments: complex (n = 16), simple-basil (n = 8), and simple-Swiss chard (n = 13). Across all three treatments, butterflies exhibited a variety of landing preferences and degrees of specialization, with some butterflies landing mostly on radish, some mostly on cabbage, and some visiting both equally (Figure 1). On average, butterflies visited 11.22 ± 0.64 (mean ± SE) cabbages in 20 host landings. A butterfly’s landing preference for cabbage, however, was not constant within an individual foraging bout; there was a higher probability that a butterfly would land on cabbage during its second ten landings than during its first 10 landings (β = 0.29 ± 0.13 SE, P = 0.03). On average, butterflies landed on 4.86 ± 0.36 (mean ± SE) cabbages during the first 10 landings and 6.35 ± 0.36 cabbages during the second 10 landings (Figure 2). Analysis using logistic regression yielded similar results; butterflies were more likely to land on cabbage as the number of host landings they made increased (β = 0.07 ± 0.01 SE, P < 0.01). Figure 1 View largeDownload slide Distribution of the number of landings on cabbage by individual butterflies during a single foraging bout of 20 host landings. A score of 0 implies that all 20 of a butterfly’s host landings were on radish; a score of 20 implies that all landings were on cabbage. On average, butterflies (n = 37) landed on 11.22 ± 0.11 (mean ± SE) cabbages in a foraging bout, indicated by the dashed line. Figure 1 View largeDownload slide Distribution of the number of landings on cabbage by individual butterflies during a single foraging bout of 20 host landings. A score of 0 implies that all 20 of a butterfly’s host landings were on radish; a score of 20 implies that all landings were on cabbage. On average, butterflies (n = 37) landed on 11.22 ± 0.11 (mean ± SE) cabbages in a foraging bout, indicated by the dashed line. Figure 2 View largeDownload slide Total cabbage landings by individual butterflies in the first period (Landings 1–10) and second period (Landings 11–20) of a foraging bout. Thin lines connect data points from the same individual butterfly and are ‘jittered’ for easier visualization. Thick black line and error bars denote the mean cabbage landings ±1.96 SE of all butterflies (n = 37) in the first and second periods. Overall, butterflies increased their landings on cabbage from the first period to the second period. Figure 2 View largeDownload slide Total cabbage landings by individual butterflies in the first period (Landings 1–10) and second period (Landings 11–20) of a foraging bout. Thin lines connect data points from the same individual butterfly and are ‘jittered’ for easier visualization. Thick black line and error bars denote the mean cabbage landings ±1.96 SE of all butterflies (n = 37) in the first and second periods. Overall, butterflies increased their landings on cabbage from the first period to the second period. Variation in the number of nonhost landings (accuracy) was best predicted by models containing host landing preference (number of cabbage landings; Table 1). Although the highest ranked model contained both host landing preference and nonhost complexity treatment (Akaike Weight = 0.51; Table 1) it was not distinguishable from a model containing only host landing preference (ΔAIC = 0.36; Table 1). Both of these models had Akaike Weights more than 5 points higher than that of the next best model and more than 7 points higher than the null model (Table 1). Analysis of the model with the lowest AICc value, containing both host landing preference and nonhost complexity, is reported in Table 2. Host landing preference was strongly associated with accuracy; butterflies that visited more cabbages also tended to visit fewer nonhosts (Table 2; Figure 3). Nonhost complexity was not a good predictor of accuracy (Table 2, Figure 4). Models of accuracy containing “landing preference” consistently outranked those containing “specialization,” indicating that a butterfly’s degree of specialization was a poor predictor of accuracy (Table 1). Table 1 Comparison of models predicting the accuracy (number of nonhost landings for every 20 host landings) of individual searching butterflies   Akaike Weights  deltaAIC  AICc  LL ratio  Nonhost Landings  Landing Pref. + Nonhost Complexity  0.51  0.00  185.57  −88.16  Landing Preference  0.43  0.36  185.93  −90.79  Landing Pref. × Nonhost Complexity  0.03  5.43  191.00  −88.10  ~ 1 NULL  0.01  7.31  192.88  −95.38  Nonhost Complexity  0.01  8.95  194.52  −93.90  Specialization  0.01  9.18  194.75  −95.20  Specialization + Nonhost Complexity  0.00  10.67  196.24  −93.49  Specialization × Nonhost Complexity  0.00  14.85  200.42  −92.81    Akaike Weights  deltaAIC  AICc  LL ratio  Nonhost Landings  Landing Pref. + Nonhost Complexity  0.51  0.00  185.57  −88.16  Landing Preference  0.43  0.36  185.93  −90.79  Landing Pref. × Nonhost Complexity  0.03  5.43  191.00  −88.10  ~ 1 NULL  0.01  7.31  192.88  −95.38  Nonhost Complexity  0.01  8.95  194.52  −93.90  Specialization  0.01  9.18  194.75  −95.20  Specialization + Nonhost Complexity  0.00  10.67  196.24  −93.49  Specialization × Nonhost Complexity  0.00  14.85  200.42  −92.81  A model with landing preference (total landings on cabbage) and nonhost complexity (diversity of nonhosts) was one of the best models predicting the accuracy of searching butterflies. View Large Table 2 Results of the best model predicting the accuracy (measured by number of nonhost landings for every 20 host landings) of individual searching butterflies   β  SE  z  P value  Nonhost Landings  Landing Preference   Total Number of Cabbage Landings  −0.04  0.01  −3.42  <0.001*  Nonhost Complexity   Simple Basil–Complex  −0.26  0.14  −1.86  0.15   Simple Swiss Chard–Complex  0.06  0.11  0.50  0.87   Simple Swiss Chard–Simple Basil  0.32  0.15  2.18  0.07    β  SE  z  P value  Nonhost Landings  Landing Preference   Total Number of Cabbage Landings  −0.04  0.01  −3.42  <0.001*  Nonhost Complexity   Simple Basil–Complex  −0.26  0.14  −1.86  0.15   Simple Swiss Chard–Complex  0.06  0.11  0.50  0.87   Simple Swiss Chard–Simple Basil  0.32  0.15  2.18  0.07  View Large Figure 3 View largeDownload slide The association between the total number of cabbage landings and the total number of nonhost landings made by individual butterflies. Each point represents data from the foraging trial of one individual butterfly, “jittered” for easier visualization. Line represents predicted values from the best-fit model. Butterflies that preferred cabbage over radish tended to land on fewer nonhosts (β = −0.04, P < 0.001). Figure 3 View largeDownload slide The association between the total number of cabbage landings and the total number of nonhost landings made by individual butterflies. Each point represents data from the foraging trial of one individual butterfly, “jittered” for easier visualization. Line represents predicted values from the best-fit model. Butterflies that preferred cabbage over radish tended to land on fewer nonhosts (β = −0.04, P < 0.001). Figure 4 View largeDownload slide The total number of nonhost landings made by individual butterflies while searching for host plants in simple (1 nonhost species) and complex (4 nonhost species) communities. Nonhost community complexity had no effect on butterfly accuracy. Bars represent means ± 1.96 SE. Figure 4 View largeDownload slide The total number of nonhost landings made by individual butterflies while searching for host plants in simple (1 nonhost species) and complex (4 nonhost species) communities. Nonhost community complexity had no effect on butterfly accuracy. Bars represent means ± 1.96 SE. Butterflies foraging in the complex nonhost environment did not specialize more than butterflies foraging in the simple basil environment (β = −0.06, P = 0.23) or butterflies foraging in the simple Swiss chard environment (β = −0.03, P = 0.55). Similarly, butterflies foraging in the complex nonhost environment had the same probability of landing on cabbages as butterflies foraging in the simple basil environment (β = −0.31, P = 0.11) and the simple Swiss chard environment (β = 0.07, P = 0. 69). A butterfly’s total time spent searching during a trial did not change with nonhost complexity (χ2 = 0.99, P = 0.61) and was not predicted by foraging accuracy (χ2 < 0.01, P = 0.99) or their interaction (χ2 = 2.52, P = 0.28). A total of 262 larvae were raised on cabbage plants (n = 130) and radish plants (n = 132). On average, larvae grew at a rate of 7.57 ± 0.12 mg/day (mean ± SE) when raised on cabbage plants and at a rate of 9.12 ± 0.12 mg/day when raised on radish plants. In our model, host identity had a significant effect on growth rate where larvae developed 1.54 ± 0.18 mg (mean ± SE) faster per day on radish than on cabbage (t = 8.77, P < 0.01). The reflectance spectra of host plants (cabbage and radish) and nonhost plants (basil, Swiss chard, lettuce, and pepper) were visually compared (Figure 5) from 300 to 700 nm, the spectral sensitivity range of the cabbage white butterfly (Shimohigashi and Tominaga 1991). The cabbage plants reflect light in the UV spectrum (300–400 nm) whereas the radish plants and other nonhost plants do not. Figure 5 View largeDownload slide The reflectance spectra of cabbage and radish host leaves (dark gray) compared to nonhost leaves (light gray). The reflectance spectrum of cabbage extends into the UV range (300–400 nm) whereas the spectra of radish and nonhost plants (pepper, Swiss chard, basil, lettuce) do not. Lines and ribbons represent the mean ± SE calculated from four individual plants of each species. Figure 5 View largeDownload slide The reflectance spectra of cabbage and radish host leaves (dark gray) compared to nonhost leaves (light gray). The reflectance spectrum of cabbage extends into the UV range (300–400 nm) whereas the spectra of radish and nonhost plants (pepper, Swiss chard, basil, lettuce) do not. Lines and ribbons represent the mean ± SE calculated from four individual plants of each species. DISCUSSION Selective information processing as a mechanism of individual specialization If selective information processing drives individual specialization, we expected the more specialized individuals to be more accurate than comparatively generalized individuals. In our study, however, only certain individuals that visited one resource were more accurate than those that visited two. Only the butterflies that primarily visited cabbage landed on fewer nonhosts than butterflies that landed on a mix of both hosts; butterflies that primarily visited radish displayed no such foraging advantage. Furthermore, despite variation in individual landing preference for cabbage vs. radish, all butterflies increased their landing preference for cabbage over time. These results offer mixed support for the hypothesis that information-processing limitations underlie patterns of individual specialization. Specialization was not associated with foraging advantages for all the resources in our assay perhaps because visiting one resource may only be advantageous if it is cues are easy to distinguish from nonresources or from the background environment. In many systems, when background contrast negatively affects the detectability of one of the target resources, another target becomes preferred (Forrest and Thomson 2009; Muchhala and Serrano 2015). In our system, cabbage may be the easier of the 2 “targets” (radish, cabbage) to distinguish from the many “distractors” (swiss chard, pepper, basil, lettuce). Cabbage leaves are covered in a wax that reflects UV wavelengths, making them visually distinct from radish and from the distractor nonhost plants (Figure 5). Cabbage white butterflies possess photoreceptors that are sensitive to these wavelengths (Shimohigashi and Tominaga 1991) and rely heavily on vision to process their environment (Snell-Rood et al. 2009) and make landing decisions (Tsuji and Coe 2014). We quantified the distinctiveness of cabbage compared to other plants in terms of vision, but it is also possible that cabbage is distinct in a completely different sensory modality given that butterflies use a combination of vision, olfaction, and chemosensation to locate hosts (Hern et al. 1996). Alternatively, the cues associated with cabbage might be attractive to butterflies not because the cues are distinguishable, but because of a tighter co-evolutionary relationship with the genus Brassica (cabbage) than with the genus Raphanus (radish), resulting in a sensory or learning bias associated with Brassica cues. The evolution of such a sensory or learning bias would be particularly expected if female landing preferences were correlated with larval performance (Gripenberg et al. 2010). However, in our greenhouse system, female cabbage whites, on average, preferred to oviposit on cabbage plants even though their larvae grow faster on radish plants. The observation that females prefer to land on a plant that is apparently less suitable for larval growth suggests the possibility of a tradeoff between larval growth rate on a host and other ecological factors related to host plant choice (presence of natural enemies, competition) or factors related to adult foraging choices (host relative abundance, host detectability/distinguishability). Alternatively, cabbage white preference for Brassica (cabbage) over Raphanus (radish) might reflect a past selection bias, if, for example, Brassica were more nutritious than Raphanus prior to their cultivation for agricultural purposes. Adult female preference for the lower quality host in our system may also be a product of temporal or spatial fluctuations in host quality in the field that we have not considered (Cronin et al. 2001) or because the cues associated with cabbage are indicators of quality in another context (e.g. nectar foraging, mate choice). Although our study suggests that information processing and specialization are related in certain cases, this study does not provide evidence that selective information processing promotes different specializations among different individuals. In the individual specialization literature, researchers are interested in the idea that individuals specialize on different resources from each other, enhancing the heterogeneity of the community (Bolnick et al. 2003; Bolnick et al. 2011). In our study, although we see variation in host landing preference among individuals during the first 10 host landings, all butterflies increased their landing preference for cabbage during the second ten host landings (Figure 2). If all individuals are becoming specialists on the same resource over time, a more homogenous community will emerge rather than the heterogeneous community that the individual specialization literature anticipates. Limits to information processing may indeed promote specialization but might not have the effect of diversifying resource use within a population, particularly if specializations are influenced by sensory or learning biases. Different ecological and evolutionary dynamics would result if specializing led to homogenous resource use rather than heterogeneous resource use among individuals (Bolnick et al. 2003). Implications of information processing in complex environments In this study, the complex environment did not appear to give individuals that visited one resource an advantage over individuals that visited two resources nor did the complex environment cause butterflies to be more specialized than butterflies in the simple environment. These results suggest several possibilities: 1) our complexity manipulation was imperceptible to the butterflies, 2) foraging primarily on one resource did not alleviate the cognitive challenges of complex nonhost environments, or 3) butterflies compensated for the difficulty of complex environments by changing the speed with which they made decisions. Unfortunately, the limitations of our experimental design do not allow us to definitively distinguish among these 3 scenarios. Although it is tempting to conclude that nonhost complexity had no effect on foraging performance in our experiment, our design has limitations that make negative results difficult to interpret. We expected that increasing the diversity of nonhost plants would increase the diversity of leaf colors, odors, shapes, and chemistries in the environment, thereby increasing the overall stimuli that a butterfly would need to process while searching. Although butterflies do use all of these cues to recognize host plants (color: Traynier 1984, Snell-Rood and Papaj 2009; odor: Ikeura, Kobayashi, and Hayata 2010; shape: Papaj 1986), we cannot explicitly confirm that the cabbage whites perceived each of the nonhosts (basil, Swiss chard, lettuce, pepper) as different from each other. Furthermore, given that the challenges of a generalist lie in keeping track of multiple types of host plants (rather than multiple types of nonhost plants), it is also possible that our nonhost manipulation had no effect on an individual’s tendency to specialize because specialization is most strongly driven by the relative abundance and diversity of host plants in the environment. Secondly, we used only 1 type of complex array (with one set of nonhost species) and only 2 types of simple arrays in our experiment. Using a broader range of both complex and simple arrays might help us see a clearer effect of complexity on search behavior. Our observation that butterflies searching in one of the simple arrays (basil) landed on marginally fewer nonhosts than butterflies searching in the other simple array (Swiss chard) suggests that the identity of the nonhost community does have the potential to influence foraging performance. Perhaps, over a wider range of simple and complex nonhost compositions involving many different nonhost species, we might have found an effect of complexity that was not observed for the limited range of nonhost compositions used. Butterflies might have compensated for the difficulty of the complex environment by taking a longer time to make decisions, as predicted by a speed-accuracy tradeoff. This tradeoff is often more prominent in difficult decision-making environments (Dyer and Chittka 2004). However, the total search time required for a butterfly to make 20 host landings was similar across all three foraging environments and unrelated to a butterfly’s accuracy. Our measure of search time, though, should be taken with some caution. Our measure included not only a butterfly’s flight time between resources, but also the time required to lay an egg, which varies greatly among individuals and is not directly related to a pre-landing decision time. Furthermore, because butterflies are ectotherms, brief cloud cover or temperature shifts during a trial can have effects on foraging speed that are independent of the difficulty of the search task. Thus, butterflies may still be compensating for complexity by changing decision speed in ways that the observer could not detect. Although butterflies might not have been able to perceive our complexity manipulation, there is evidence from the literature to suggest that the diversity of distractors in a search task may not influence foraging efficiency even when searchers can perceive the change in complexity. For example, bees foraging for targets among only blue distractors had the same error rate and decision time as bees foraging for the same targets among mixed-color distractors (Spaethe et al. 2006). Kostenko et al. (2015) found that the diversity of nonhost plants in a field experiment did not influence a parasitoid’s host location success but that structural complexity (no vegetation, mowed vegetation, tall vegetation) did influence aggregation behavior. In the agricultural literature, planting a diversity of companion plants with a target crop often has variable results on pest populations (Andow 1991; for Brassicaeceae crops specifically, Hooks and Johnson 2003). It is difficult to draw direct conclusions about the effect of nonhost plant diversity on the insect search process from many field studies because manipulating companion crop diversity alters other confounding factors like floral resources, natural enemies, and belowground plant competition (Letourneau et al. 2011). Our study, which controlled for all of these factors, suggests that, in some cases, the diversity of nonhosts may not hamper an insect’s ability to locate hosts. There may be other features of the distractor community that have a greater effect on decision-making than diversity. For a sequentially foraging animal like the cabbage white butterfly, decision-making in a series is more likely to be influenced by the total number of times the searcher encounters distractors in a sequence (influenced by the ratio of distractors to targets; Treisman and Gelade 1980) rather than the different kinds of distractors (influenced by diversity). When bees were trained to distinguish between rewarding target flowers and nonrewarding distractors, error rate increased 4-fold and decision time increased 1.5-fold as the number of distractor flowers in the assay increased from one to fifteen (Spaethe, Tautz, and Chittka 2006). The identity of distractors may also influence the difficulty of a foraging task. When a series of 24 distractor nonhost plants were intercropped (each individually) with a target host Brassica plant, the number of eggs laid by the cabbage root fly was affected differently by each type of distractor (Finch et al. 2003). A variety of mechanisms might make certain nonhost plants more effective at masking host plants than others. For example, searching might be more difficult when distractor properties make discriminating between targets and distractors more difficult. The error rate of foraging bees increased when target and distractor flowers had similar colors (Dyer and Chittka 2004). This idea that distractor identity influences foraging accuracy is partially supported in our data. Butterflies foraging in the basil nonhost environment were marginally (though not significantly) more accurate than butterflies foraging in our Swiss chard nonhost environment (Table 2; Figure 4). Considering that basil has a distinct olfactory profile, it might have been easier for butterflies to discriminate between cabbage/radish targets and very obvious basil distractors than between cabbage/radish targets and Swiss chard distractors. These observations suggest nonhost identity might create cognitive challenges for the butterflies, possibly resulting in individual specialization. CONCLUSIONS This work provides partial support for the idea that selective attention drives individual specialization because certain individuals that visited 1 resource (cabbage) foraged more accurately than individuals that visited 2 resources (cabbage and radish). However, because not all resources provided the foraging benefits associated with specialization, selective attention may not drive individuals to develop specializations that are different from each other. If we want to further understand how behavioral mechanisms (like learning or information processing) influence individual specialization on various timescales, experiments that manipulate both a forager’s experience and environment over time will be particularly important. Such detailed individual-level data, although difficult to obtain, are necessary to understand how specialized resource preferences develop, how long they last, and how they interact with environmental context. Further clarifying the mechanisms behind individual specialization will allow us to make concrete predictions about the environmental factors that drive specialization and the degree of resource use heterogeneity among individuals in a population. FUNDING This work was supported by the Bell Museum of Natural History (Dayton Bell Museum Fund Fellowship). The Snell-Rood lab was supported in part through NSF-IOS-1354737 during this time. Data accessibility Analyses reported in this article can be reproduced using the data provided by Steck and Snell-Rood (2017). Many thanks to Kush Patel, Monika Malek, Rhea Smykalski, Chase Marx, and Regina Kurandina for help collecting butterflies, caring for them, rearing larvae, and growing plants. We appreciated discussion and feedback from the Snell-Rood lab and from post-docs Eli Swanson and Eric Lind. We also thank three anonymous reviewers and the handling editor for very thoughtful, constructive comments that greatly improved the quality of this manuscript. REFERENCES Andow D. 1991. Vegetational diversity and arthropod population response. Annu Rev Entomol . 36: 561– 586. Google Scholar CrossRef Search ADS   Araújo MS, Bolnick DI, Layman CA. 2011. The ecological causes of individual specialisation. Ecol Lett . 14: 948– 58. Google Scholar CrossRef Search ADS PubMed  Bates D, Maechler M, Bolker B, Walker S. 2014. lme4: Linear mixed-effects models using Eigen and S4. R package version1.1–7. Available from: http://CRAN.Rproject.org/package=lme4. Bauerfiend SS, Fischer K. 2008. Maternal body size as a morphological constraint on egg size and fecundity in butterflies. Bas Appl Ecol . 9: 443– 451. Google Scholar CrossRef Search ADS   Bernays EA, Wcislo WT. 1994. Sensory capabilities information processing and resource specialization. Q Rev Biol . 69: 187– 204. Google Scholar CrossRef Search ADS   Bernays EA. 1997. Feeding by lepidopteran larvae is dangerous. Ecol Entomol . 22: 121– 123. Google Scholar CrossRef Search ADS   Bernays EA. 1998. The value of being a resource specialist: behavioral support for a neural hypothesis. Am Nat . 151: 451– 464. Google Scholar CrossRef Search ADS PubMed  Bernays EA. 2001. Neural limitations in phytophagous insects : implications for diet breadth and evolution of host affiliation. Annu Rev Entomol . 46: 703– 727. Google Scholar CrossRef Search ADS PubMed  Bernays EA, Singer MS, Rodrigues D. 2004. Foraging in nature: Foraging efficiency and attentiveness in caterpillars with different diet breadths. Ecol Entomol . 29: 389– 397. Google Scholar CrossRef Search ADS   Bolnick DI, Yang LH, Fordyce JA, Davis JM, Svanbäck R. 2002. Measuring individual-level resource specialization. Ecology . 83: 2936– 2941. Google Scholar CrossRef Search ADS   Bolnick DI, Svanbäck R, Fordyce JA, Yang LH, Davis JM, Hulsey CD, Forister ML. 2003. The ecology of individuals: incidence and implications of individual specialization. Am Nat . 161: 1– 28. Google Scholar CrossRef Search ADS PubMed  Bolnick DI, Amarasekare P, Araújo MS, Bürger R, Levine JM, Novak M, Rudolf VHW, Schrieber SJ, Urban MC, Vasseur DA. 2011. Why intraspecific trait variation matters in community ecology. Trends Ecol Evol . 26: 183– 192. Google Scholar CrossRef Search ADS PubMed  Brosi BJ. 2016. Pollinator specialization: from the individual to the community. New Phytol . 210: 1190– 1194. Google Scholar CrossRef Search ADS PubMed  Chittka L, Skorupski P, Raine NE. 2009. Speed-accuracy tradeoffs in animal decision making. Trends Ecol Evol . 34: 400– 407. Google Scholar CrossRef Search ADS   Cronin JT, Abrahamson WG, Craig TP. 2001. Temporal variation in herbivore host-plant preference and performance: constraints on host-plant adaptation. Oikos . 93: 312– 320. Google Scholar CrossRef Search ADS   Dall SRX, Cuthill IC. 1997. The information costs of generalism. Oikos . 80: 197– 202. Google Scholar CrossRef Search ADS   Dall SRX, Bell AM, Bolnick DI, Ratnieks FLW. 2012. An evolutionary ecology of individual differences. Ecol Lett . 15: 1189– 1198. Google Scholar CrossRef Search ADS PubMed  Darimont CT, Paquet PC, Reimchen TE. 2007. Stable isotopic niche predicts fitness of prey in a wolf-deer system. Biol J Linn Soc . 90: 125– 137. Google Scholar CrossRef Search ADS   Dieckmann U, Doebeli M. 1999. On the origin of species by sympatric speciation. Nature . 400: 354– 357. Google Scholar CrossRef Search ADS PubMed  Duffy MA. 2010. Ecological consequences of intraspecific variation in lake Daphnia. Freshwater Biol . 55: 995– 1004. Google Scholar CrossRef Search ADS   Dukas R, Kamil AC. 2001. Limited attention : the constraint underlying search image. Behav Ecol . 12: 192– 199. Google Scholar CrossRef Search ADS   Dukas R. 2002. Behavioural and ecological consequences of limited attention. Philos Trans R Soc Lond B Biol Sci . 357: 1539– 1547. Google Scholar CrossRef Search ADS PubMed  Dukas R. 2004. Causes and consequences of limited attention. Brain Behav Evol . 63: 197– 210. Google Scholar CrossRef Search ADS PubMed  Dusenbery DB. 1992. Sensory ecology: how organisms acquire and respond to information . New York (NY): W.H. Freeman. Dyer AG, Chittka L. 2004. Biological significance of distinguishing between similar colours in spectrally variable illumination: Bumblebees (Bombus terrestris) as a case study. J. Comp. Physiol. A Neuroethol. Sensory, Neural, Behav. Physiol . 190: 105– 114. Google Scholar CrossRef Search ADS   Egan SP, Funk DJ. 2006. Individual advantages to ecological specialization: insights on cognitive constraints from three conspecific taxa. Proc Biol Sci . 273: 843– 848. Google Scholar CrossRef Search ADS PubMed  Estes JA, Riedman ML, Staedler MM, Tinker MT, Lyon BE. 2003. Individual variation in prey selection by sea otters: patterns, causes and implications. J Anim Ecol . 72: 144– 155. Google Scholar CrossRef Search ADS   Finch S, Billiald H, Collier RH. 2003. Companion planting - do aromatic plants disrupt host-plant finding by the cabbage root fly and the onion fly more effectively than non-aromatic plants? Entomol Exp Appl . 109: 183– 195. Google Scholar CrossRef Search ADS   Forrest J, Thomson JD. 2009. Background complexity affects colour preference in bumblebees. Naturwissenschaften  96: 921– 925. Google Scholar CrossRef Search ADS PubMed  Fox LR, Morrow PA. 1981. Specialization: species property or local phenomenon? Science . 211: 887– 893. Google Scholar CrossRef Search ADS PubMed  Fox J, Weisberg S. 2011. An {R} companion to applied regression , Second Edition. Thousand Oaks (CA): Sage. Available from: http://socserv.socsci.mcmaster.ca/jfox/Books/Companion. García-Barros E. 2000. Body size, egg size, and their interspecific relationships with ecological and life history traits in butterflies (Lepidoptera: Papilionoidea, Hesperiodiea). Biol J Linn Soc . 70: 251– 284. Google Scholar CrossRef Search ADS   Gols R, Bukovinszky T, Hemerik L, Harvey JA, Van Lenteren JC, Vet LEM. 2005. Reduced foraging efficiency of a parasitoid under habitat complexity: implications for population stability and species coexistence. J Anim Ecol . 74: 1059– 1068. Google Scholar CrossRef Search ADS   Gripenberg S, Mayhew P, Parnell M, Roslin, T. 2010. A meta-analysis of preference-performance relatinoships in phytophagous insects. Ecol Lett . 13: 383– 393. Google Scholar CrossRef Search ADS PubMed  Heinrich B. 1976. The foraging specializations of individual bumblebees. Ecol Mono . 46: 105– 128. Google Scholar CrossRef Search ADS   Heinrich B. 1993. Caterpillars: ecological and evolutionary constraints on foraging . New York (NY): Chapman & Hall. Hern A, Edwards-Jones G, McKinlay RG. 1996. A review of the pre-oviposition behaviour of the small cabbage white butterfly, Pieris rapae (Lepidoptera : Pieridae). Annu. Appl Biol  128: 349– 371. Google Scholar CrossRef Search ADS   Hooks CRR, Johnson MW. 2003. Impact of agricultural diversification on the insect community of cruciferous crops. Crop Prot . 22: 223– 238. Google Scholar CrossRef Search ADS   Hothorn T, Bretz F, Westfall P. 2008. Simultaneous inference in general parametric models. Biom J . 50: 346– 363. Google Scholar CrossRef Search ADS PubMed  Hughes AR, Inouye BD, Johnson MTJ, Underwood N, Vellend M. 2008. Ecological consequences of genetic diversity. Ecol Lett . 11: 609– 623. Google Scholar CrossRef Search ADS PubMed  Ikeura H, Kobayashi F, Hayata Y. 2010. How do Pieris rapae search for Brassicaceae host plants? Biochem Sys Ecol . 38: 1199– 1203. Google Scholar CrossRef Search ADS   Jaumann S, Snell-Rood EC. 2017. Trade-offs between fecundity and choosiness in ovipositing butterflies. Anim Behav . 123: 433– 440. Google Scholar CrossRef Search ADS   Janz N, Nylin S. 1997. The role of female search behaviour in determining host plant range in plant feeding insects: a test of the information processing hypothesis. Proc R Soc B Biol Sci . 264: 701– 707. Google Scholar CrossRef Search ADS   Janz N. 2003. The cost of polyphagy: oviposition decision time vs error rate in a butterfly. Oikos . 100: 493– 496. Google Scholar CrossRef Search ADS   Johnson CK, Tinker MT, Estes JA, Conrad PA, Staedler M, Miller MA, Jessup DA, Mazet JAK. 2009. Prey choice and habitat use drive sea otter pathogen exposure in a resource-limited coastal system. Proc Natl Acad Sci USA . 106: 2242– 2247. Google Scholar CrossRef Search ADS PubMed  Kastner S, Ungerleider LG. 2000. Mechanisms of visual attention in the human cortex. Annu Rev Neurosci . 23: 315– 341. Google Scholar CrossRef Search ADS PubMed  Kocher SD, Williams EH. 2000. The diversity and abundance of North American butterflies cary with habitat disturbance and geography. J Biogeogr . 27: 785– 794. Google Scholar CrossRef Search ADS   Kostenko O, Lammers M, Grootemaat S, Kroon T, Harvey JA, Van Geem M, Bezemer TM. 2015. Effects of plant diersity and structural complexity on parasitoid behavior in a field experiment. Ecol Entomol . 40: 748– 758. Google Scholar CrossRef Search ADS   Layman CA, Newsome SD, Gancos Crawford T, Crawford TG. 2015. Individual-level niche specialization within populations: emerging areas of study. Oecologia . 178: 1– 4. Google Scholar CrossRef Search ADS PubMed  Letourneau DK, Armbrecht I, Rivera BS, Lerma M, Carmona EJ, Daza MC, Escobar S, Galindo V, Gutierrez C, Lopez SDet al.   2011. Does plant diversity benefit agroecosystems? Ecol Appl . 21: 9– 21. Google Scholar CrossRef Search ADS PubMed  Lomnicki A. 1978. Individual differences between animals and the natural regulation of their numbers. J Anim Ecol . 47: 461– 475. Google Scholar CrossRef Search ADS   Maia R, Eliason CM, Bitton PP, Doucet SM, Shawkey MD. 2013. pavo: an R Package for the analysis, visualization and organization of spectral data. Methods Ecol Evol . 4: 906– 913. Mazerolle MJ. 2015. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 2.0–3. Available from: http://CRAN.R-project.org/package=AICcmodavg. Moran J, Desimone R. 1985. Selective attention gates visual processing in the extrastriate cortex. Science . 229: 782– 784. Google Scholar CrossRef Search ADS PubMed  Muchhala N, Serrano D. 2015. The complexity of background clutter affects nectar bat use of flower odor and shape cues. PLoS One  10: 1– 12. Google Scholar CrossRef Search ADS   Papaj D. 1986. Conditioning of leaf-shape discrimination by chemical cues in the butterfly, Battus philenor. Anim Behav . 34: 1281– 1288. Google Scholar CrossRef Search ADS   Perfecto I, Vet LEM. 2003. Effect of a non-host plant on the location behavior of two parasitoids: the tritrophic system of Cotesia spp. (Hymenoptera : Braconidae), Pieris rapae (Lepidoptera : Pieridae), and Brassica oleraceae. Environ Entomol . 32: 163– 174. Google Scholar CrossRef Search ADS   R Core Team. 2013. R: A language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Available from: http://www.R-project.org/. Raitanen J, Forsman JT, Kivelä SM, Mäenpää MI, Välimäki P. 2014. Attraction to conspecific eggs may guide oviposition site selection in a solitary insect. Behav Ecol . 25: 110– 116. Google Scholar CrossRef Search ADS   Reimchen TE, Nosil P. 2001. Dietary differences between phenotypes with symmetrical and asymmetrical pelivs in the stickleback Gasterosteus aculeatus. Can J Zool . 79: 533– 539. Google Scholar CrossRef Search ADS   Renwick J, Chew F. 1994. Oviposition behavior in lepidoptera. Annu Rev Entomol . 39: 377– 400. Google Scholar CrossRef Search ADS   Scott JA. 1986. The Butterflies of North America: A Natural History and Field Guide . Standford, California: Stanford University Press. Shimohigashi M, Tominaga Y. 1991. Identification of UV, green and red receptors, and their projection to lamina in the cabbage butterfly, Pieris rapae. Cell Tissue Res . 263: 49– 59. Google Scholar CrossRef Search ADS   Snell-Rood EC, Papaj DR. 2009. Patterns of phenotypic plasticity in common and rare environments: a study of host use and color learning in the cabbage white butterfly Pieris rapae. Am Nat . 173: 615– 631. Google Scholar CrossRef Search ADS PubMed  Snell-Rood EC, Papaj DR, Gronenberg W. 2009. Brain Size: A global or induced cost of learning? Brain Behav Evol . 73: 111– 128. Google Scholar CrossRef Search ADS PubMed  Spaethe J, Tautz J, Chittka L. 2006. Do honeybees detect colour targets using serial or parallel visual search? J Exp Biol . 209: 987– 993. Google Scholar CrossRef Search ADS PubMed  Steck MK, Snell-Rood E. 2017. Data from: specialization and accuracy of host-searching butterflies in complex and simple environments. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.s70m4. Svanbäck R, Bolnick DI. 2007. Intraspecific competition drives increased resource use diversity within a natural population. Proc Biol Sci . 274: 839– 844. Google Scholar CrossRef Search ADS PubMed  Terraube J, Arroyo B, Madders M, Mougeot F. 2011. Diet specialisation and foraging efficiency under fluctuating vole abundance: A comparison between generalist and specialist avian predators. Oikos  120: 234– 244. Google Scholar CrossRef Search ADS   Traynier RMM. 1984. Associative learning in the ovipositional behavior of the cabbage white butterfly, Pieris rapae. Phys Entomol . 9: 465– 472. Google Scholar CrossRef Search ADS   Treisman AM, Gelade G. 1980. A feature-integration theory of attention. Cogn Psychol . 12: 97– 136. Google Scholar CrossRef Search ADS PubMed  Tsuji J, Coe L. 2014. Effects of foliage color on the landing response of Pieris rapae (Lepidoptera: Pieridae). Environ Entomol . 43: 989– 994. Google Scholar CrossRef Search ADS PubMed  Van Valen L. 1965. Morphological variation and width of ecological niche. Am Nat . 99: 377– 390. Google Scholar CrossRef Search ADS   Wee B, Singer M. 2007. Variation among individual butterflies along a generalist – specialist axis : no support for the “ neural constraint “ hypothesis. Ecol Entomol . 32: 257– 261. Google Scholar CrossRef Search ADS   Xiao F, Cuthill IC. 2016. Background complexity and the detectability of camouflaged targets by birds and humans. Proc R Soc B . 283: 20161527. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral Ecology Oxford University Press

Specialization and accuracy of host-searching butterflies in complex and simple environments

Loading next page...
 
/lp/ou_press/specialization-and-accuracy-of-host-searching-butterflies-in-complex-pFlZTyq3Nx
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1045-2249
eISSN
1465-7279
D.O.I.
10.1093/beheco/ary001
Publisher site
See Article on Publisher Site

Abstract

Abstract Populations that have access to a variety of resources are often composed of individuals that specialize on different subsets of resources. Understanding the behavioral mechanisms that drive such individual specialization will help us predict the strength of this specialization across different environments. Here, we explore the idea that individual specialization may be a consequence of constraints on an individual’s ability to process information. Because many environments contain an overwhelming number of resources and associated stimuli, individuals that specialize by focusing on only a subset of these resources may make more accurate decisions than individuals that generalize. Furthermore, we expect individuals in complex environments, where there are more resources and associated stimuli to process, to specialize during their search for resources compared with individuals in simple environments. We tested these predictions by measuring the accuracy and degree of specialization of naïve cabbage white butterflies (Pieris rapae) searching for 2 target host species (radish and cabbage) in simple (1 distractor species) and more complex (4 distractor species) environments. Only butterflies that specialized on cabbage were more accurate than butterflies that visited a mixture of both radish and cabbage. Furthermore, naïve butterflies searching for hosts in complex environments did not adopt more specialized foraging strategies than naive butterflies searching for hosts in simple environments. Taken together, these results suggest that the foraging benefits associated with specialization might only apply to certain resources (perhaps those that have readily recognizable cues) and that such specializations can be related to accuracy across multiple environments. INTRODUCTION Individuals in a population often use only a small subset of the resources available to the entire population (Van Valen 1965; Fox and Morrow 1981; Bolnick et al. 2003). This pattern of resource use, termed individual specialization, has been observed across a variety of taxa and can have ecological and evolutionary consequences (Estes et al. 2003; Dall et al. 2012; Bolnick et al. 2011). The degree to which individuals in a population differ in their resource use can affect population dynamics and stability (Lomnicki 1978; Hughes et al. 2008), species interaction networks (e.g. pollinators in Brosi 2016), intra- and inter-specific competition (Svanbäck and Bonlick 2007; Duffy 2010), and may ultimately result in sympatric speciation (Dieckmann and Doebeli 1999). Individual specialization also has consequences for the individual organism. Specializing on only one resource can affect an individual’s foraging performance (Bernays 1998), vulnerability to predation (Darimont et al. 2007), and risk of parasite exposure (Reimchen and Nosil 2001; Johnson et al. 2009). Over the past decade, research on individual specialization has moved from documenting the presence or absence of individual specialization in nature to exploring specific hypotheses regarding variation in specialization over time and space (Araújo et al. 2011; Dall et al. 2012; Layman et al. 2015). To make more explicit predictions about how and why specialization changes across ecological gradients, we need to better understand the behavioral mechanisms that promote and facilitate individual resource specializations. From a behavioral perspective, specialization may arise from limitations on the amount of information an organism can process (Bernays 2001; Dukas 2002). The environment contains more relevant information than any one individual can actually evaluate and, consequently, many animals (including humans) must focus on specific incoming signals and ignore others in order to make choices efficiently (Dusenbery 1992; Kastner and Ungerleider 2000). When humans, monkeys, and birds are faced with the task of detecting multiple stimuli, focusing attention on one of these stimuli increases the probability of detecting that stimulus (Moran and Desimone 1985; Kastner and Ungerleider 2000; Dukas and Kamil 2001). This idea, that there are costs to divided attention, is often referred to as “neural constraints,” “limited attention,” or the “information-processing hypothesis” (Bernays 2001; Dukas 2004; Egan and Funk 2006) and has implications for how animals make resource-related choices. Because focusing on a small number of resources requires less attentional load than focusing on a large number resources, the information-processing hypothesis predicts that specialists make more efficient decisions than generalists (Bernays and Wcislo 1994; Dall and Cuthill 1997; Bernays 2001). Specializing on a subset of available resources could increase a forager’s speed and/or accuracy (ability to distinguish between rewarding and nonrewarding resources), 2 key components of efficient decision-making (Chittka et al. 2009). Field studies have lent support to the idea that specialists are more accurate than generalists by testing this prediction across species, showing that specialist species are often more accurate and efficient than generalist species (Janz and Nylin 1997; Bernays, Singer, and Rodrigues 2004; Terraube et al. 2011). However, variation in specialization may also occur among individuals within a population (Fox and Morrow 1981; Bolnick et al. 2003). For example, certain individual bumblebee workers foraging in the same area might visit only one species of flower while other individuals might visit three species of flowers (Heinrich 1976). This example of resource specialization is a flexible behavioral strategy that can vary within and among individuals as well as across ecological contexts (Brosi 2016). Constraints on information processing could contribute to this individual specialization; however, there has been less exploration as to whether specialized individuals in a population receive the same decision-making advantages we see in specialized species. The studies that do address selective information processing on the individual level present mixed support for the idea that more specialized individuals make more accurate and efficient decisions than more generalized individuals. Egan and Funk (2006) found that specialized leaf beetle “host forms” were more accurate than generalist forms and Janz (2003) found that polyphagous members of the butterfly Polygonum c-album had longer decision times than specialists. However, in the butterfly Euphydryas editha, natural variation in specialization among individuals was not associated with faster host recognition or better accuracy (Wee and Singer 2007). Given these conflicting results, our work will address whether and under what conditions selective processing of information contributes to resource specialization on the individual level. The information-processing hypothesis not only predicts consequences for specialized and generalized search strategies, it also has implications for how individuals interact with environments that vary in cognitive complexity. A complex environment can make resource decisions more challenging by decreasing the ratio of resources (called “targets”) to nonresources (called “distractors”), increasing the similarity between targets and distractors (“discriminability”) and increasing the similarity between targets/distractors and the background in which targets and distractors are embedded (“detectability”). Each of these different types of complexity can reduce the accuracy and efficiency of individuals that are searching for resources (Perfecto and Vet 2003; Gols et al. 2005; Xiao and Cuthill 2016). If resource-related decisions become more difficult in complex environments, specialists, because they have the opportunity to simplify their decisions, may have an advantage. Thus, we can make two predictions about how information processing might influence individual search behavior in complex environments. Firstly, individuals searching in complex environments might be expected to adopt a more specialized foraging strategy than different individuals searching in simple environments. Secondly, more specialized individuals are expected to make more accurate decisions than more generalized individuals. To explore these predictions, we observed individual naïve cabbage white butterflies (Pieris rapae) searching for 2 different species of host plants embedded in either complex (multiple species of nonhost plants) or simple (one species of nonhost plant) communities. In the field, cabbage white butterflies lay eggs on a variety of host plant species in the family Brassicaceae (Scott 1986). In the lab, we further observed that individual butterflies vary along a continuum where some individuals oviposit on multiple host species and others oviposit primarily on one host species, even when viable alternate hosts are available. Because cabbage whites are relatively small and display fairly typical host searching behavior in large greenhouse flight cages, we can observe the behavior of a single butterfly and calculate both the individual’s degree of host specialization when offered multiple hosts and the individual’s foraging accuracy. Although insect specialization can be measured across a variety of life stages, contexts, and timescales, we measure specialization in the present study as the tendency of an adult female butterfly to lay her eggs on primarily one host plant species (more specialized) rather than multiple host plant species (more generalized) during one bout of host searching. If limits to information processing promote individual specialization in this system, we first predict that more specialized butterflies (those that visit only one of the 2 host species) will be more accurate than more generalized butterflies (those that visit both host species). If complex environments place further strain on information processing, we expect the individual butterflies searching for hosts in complex environments (with multiple species of nonhost plants) to be more specialized than the individual butterflies searching for hosts in simple environments (with one species of nonhost plant). METHODS Care for experimental butterflies Adult Pieris rapae used in this experiment were the offspring of over forty wild-caught females captured in Saint Paul, Minnesota during Summer 2014. Larvae were reared on artificial diet (recipe in Jaumann and Snell-Rood 2017) in a 14:10 h light:day climate chamber at 23 °C and approximately 60% relative humidity. When butterflies emerged from their chrysalides, they were immediately marked with an identification number on their hind wing and stored in glassine envelopes in a refrigerator at 5 °Celsius for no more than four days. Butterflies were subsequently transferred to 60 cm3 “bug dorm” mesh cages where they were allowed to mate with males for 48 h on an ad lib diet of 10% honey water solution before behavioral testing in our experimental plant arrays. Plant care and measurements Cabbage (Brassica oleraceae var. Earliana) and radish (Raphanus sativus var. Cherry Belle) were selected as host plants because both Brassica and Raphanus species are commonly planted on farms, can be found in organic gardens, and have related feral species commonly used by the cabbage white butterfly (Scott 1986). Radish and cabbage have relatively similar growing times and similar leaf area at 40 days of growth. Basil, Swiss chard, lettuce and pepper (basil: Ocimum basilicum, Swiss chard: Beta vulgaris var. Fordhook®Giant, lettuce: Lactuca sativa var. Nevada, pepper Capsicum annuum var. California Wonder) were selected as nonhost plants because they were similar in size and growing time to hosts but added a wide variety of leaf colors, shapes, and smells to environment. These nonhost plants are also commonly found in agricultural settings and would likely be present in the human-dominated disturbed landscapes where cabbage whites are most abundant (Kocher and Williams 2000). All flowers and buds on plants were removed and tall nonhost plants were cut to match host plant size to standardize structural complexity across trials. Seeds were first sown in May 2014, and new seeds were germinated every three weeks through late July. Seeds were germinated in pit trays and transplanted after 2 weeks to (5” × 5”) 2.65-liter pots filled with Sungro® potting mix (70–80% Sphagnum). One week after transplanting, all plants were fertilized with 6 g of 14-14-14 Osmocote® fertilizer and treated with OHP Marathon 1% Granular (a systemic pesticide with 1% Imidacloprid) to control greenhouse pest damage. Plants were kept in a greenhouse and watered daily. To compare color differences among our focal plants, 4 plants of each species were haphazardly selected for spectral analysis with an Ocean Optics Inc. JAZ-PX spectrometer. We calculated the mean reflectance and standard error of each species from 300 to 700 nm, the spectral sensitivity range of the cabbage white butterfly (Shimohigashi and Tominaga 1991). Reflectance was graphed with the “pavo” package in R, using a smoothing spectra with a span of 0.2 (Maia et al. 2013). Experimental arrays Adult cabbage white butterflies were allowed to search for host plants in one of 3 treatment arrays that varied in community complexity (one complex environment with 4 nonhosts and 2 simple environments, each with a single different nonhost species). Each treatment array contained a total of 24 potted plants, 12 of which were host plants (6 cabbage, 6 radish) and 12 of which were nonhost plants. These potted plants were arranged in a 6 × 4 rectangular array approximately 12 cm apart with no overlapping foliage inside a 1.6 × 2m flight cage. The number and identity of host plants (6 radish, 6 cabbage) remained constant over all 3 treatment arrays. Potted host plants were placed such that all immediate neighboring pots were of a different species. Adjacent neighbors were always nonhosts and diagonal neighbors were hosts of the other species. This design was chosen in an effort to minimize the patchiness inherent in the haphazard placement of plants. To manipulate environmental complexity, we changed the identity of the 12 nonhosts in each of the three treatment arrays. The “complex” treatment contained 4 different nonhost species (3 plants each of basil, Swiss chard, lettuce, and pepper) whereas the 2 “simple” treatments contained only one nonhost species (12 basil plants and 12 Swiss chard plants respectively). Nonhosts were placed such that members of the same species were as far apart from each other as possible. We chose to manipulate complexity by increasing nonhost diversity because such a manipulation increases the overall diversity of plants (and associated stimuli) in the environment without changing the host to nonhost ratio, the relative frequency of host plants, or the identity of host plants. Behavioral observations Butterflies were randomly assigned to a complexity treatment and tested one at a time by placing a focal individual on a randomly assigned host plant, in order to motivate host searching. Butterflies that did not immediately begin searching were not tested. The butterflies that immediately began searching were observed until they completed 20 host landings (landings on either cabbage or radish). Butterfly “landings” were characterized by tarsal drumming on the leaf surface, a behavior associated with chemoreception using tarsal receptors (Renwick and Chew 1994). Other behaviors during which butterflies were in contact with the plants (such as basking or resting) could be distinguished from “landings” because they are not accompanied by drumming and are marked by prolonged still behavior with wings held closed or slightly open at a 45 degree angle. After a host plant landing, females often lay a single egg and then continue to search for plants. Landings on both cabbage and radish result in an egg being laid 87.1% and 87.3% of the time respectively (for Ncabbage landings = 582 and Nradish landings = 458, 2015 data). One observer (MS) recorded landings on host and nonhost plants, the species of the plant, and the occurrence of drumming and oviposition. Re-visits to plants were counted as a new landing if butterflies flew over a neighboring plant in between. Each behavior and the time it occurred during the trial were recorded on JWatcher software installed on a laptop computer. No more than four butterflies were tested sequentially per treatment and the order of the treatments was changed on a daily basis. The eggs laid by each butterfly were gently brushed off the host plants after every trial to minimize the searching butterfly’s use of conspecific cues (Raitanen et al. 2014). Out of 96 butterflies that were initially released into the assay, a total of 37 butterflies successfully completed 20 host landings and were retained for analysis. Observations took place in a flight cage (2 × 1.6 × 1.6 m high PVC frame covered with nontreated mosquito netting) in the plant growth facilities greenhouses at the University of Minnesota between the hours of 10 am and 3 pm on sunny days from 30 June 2014 to 15 August 2014. Behavioral analyses Each female’s behavior was analyzed over her 20 host landings to calculate total host specialization, host landing preference, and host-finding accuracy. Host specialization and host landing preference were both calculated from a butterfly’s landings on host plants and served to situate each individual along a continuous gradient of resource use rather than as a binary “generalist” or “specialist.” However, host specialization and host landing preference each measure a slightly different aspect of host use. “Host specialization” measures a butterfly’s tendency to land primarily on one host species (regardless of whether this species is cabbage or radish) compared to two host species, spanning a gradient where more specialized individuals (landing on one species) represent one extreme and more generalized individuals (landing on 2 species) represent the other. Host specialization was estimated by calculating how far a butterfly’s landings on cabbage and radish departs from the expected number of landings given the number of cabbages in the assay (represented as 0.5 in the following equation).  host specialization = | (landings on cabbage) / (20 landings) − 0.5 | This metric ranges from 0 to 0.5, where 0.5 indicates that a butterfly landed either 20 times on cabbage or 20 times on radish, and 0 indicates that a butterfly landed on 10 cabbages and 10 radishes. There are many indices recommended for quantifying individual specialization that focus on how specialized an individual’s resource use is compared to the resource use of the population (outlined in Bolnick et al. 2002). We choose to follow the formula above because we want to focus on a measure of specialization that calculates how far an individual departs from resource availability in the controlled experimental environment. In contrast, “host landing preference” measures a butterfly’s tendency to land on cabbage compared to radish, a gradient where individuals that land primarily on cabbage are on one extreme and individuals that land primarily on radish are on the other, with the more generalized individuals falling in the middle of this axis. Host landing preference was estimated by counting the total number of cabbage landings out of 20 host landings, leading to a score from 0 to 20 where 0 indicates that all 20 host landings were made on radish and 20 indicates that all host landings were made on cabbage. We quantified a butterfly’s host-finding accuracy by counting the number “inaccurate” landings a butterfly made, i.e. the number nonhost plants a female drummed during the course of a foraging trial. Since all butterflies were allowed to make the same number of host landings and often laid eggs immediately following these landings, we assume that each butterfly had the opportunity to lay a similar number of eggs and that variation in search performance among butterflies is due largely to nonhost landings. To the extent that nonhost landings increase the time spent host searching (at the expense of other activities), and decrease the number of eggs laid during an oviposition bout, butterflies that land on more nonhosts would be expected to lay fewer eggs over their lifetime (Snell-Rood and Papaj 2009). To approximate the each butterfly’s total search time, we isolated the portions of a trial during which a butterfly was actively searching for or interacting with host plants (i.e. only drumming and oviposition behaviors from our ethogram) and added these portions together. Larval performance To estimate the consequences of a female’s oviposition choice, we raised cabbage white larvae on an ad lib diet of entire radish plants or cabbage plants and calculated their growth rate (pupal mass/development time). Gaining mass and developing quickly are considered advantages for butterflies because body size is often positively correlated with fecundity across species (García-Barros 2000; but see Bauerfiend and Fischer 2008 for a discussion of variation within species) and short development times allow larvae to minimize their vulnerability to predators while feeding (Heinrich 1993, Bernays 1997). To collect eggs for rearing, we allowed 10 wild-caught butterflies to oviposit on cabbage in one cage and 10 different butterflies to oviposit on radish in another cage, for 6 h. The plants with eggs were kept in these cages until they hatched and the larvae were large enough to transfer without injury (about 7 days after laying). Larvae were then distributed to new cages with fresh cabbage plants and fresh radish plants respectively so that there were 20 larvae per cage. The plants used for larval rearing were grown according to the same protocol as plants for behavioral trials except that no insecticide was applied. New plants were supplied to all rearing cages as needed over the course of larval development to provide an ad lib diet. The cages were monitored daily and once caterpillars formed pupae, they were collected and weighed on a microbalance. To calculate development time, we counted the number of days from the date the egg was laid to the date the larvae was collected. Growth rate was calculated for each larva by dividing its pupal weight by its development time. We repeated this experiment in four rounds throughout the month of August 2015. Statistics We used generalized linear regression models (GLM) for all of our analyses using R version 3.0.2 (R Core Team 2013). To test whether specialist butterflies were more accurate than generalist butterflies, we modeled how accuracy (nonhost landings) was affected by either specialization or landing preference (respectively) and nonhost community complexity. Specialization and landing preference were never used in the same model because they are both metrics calculated from the same raw data: a butterfly’s landings on cabbage versus radish. Accuracy was modeled using a Poisson distribution because nonhost landings were counted within a defined behavioral set (20 host landings) and had no upper bound. Because there could feasibly be interactions between nonhost complexity and specialization/landing preference on nonhost landings, we began with the full model and used corrected Akaike information criteria (AICc) and Akaike weights to compare the full model to simpler models (“AICcmodavg” package in R; Mazerolle 2015). We ran Tukey’s pairwise contrasts on the best-ranked model to identify differences between complexity treatments (“multcomp” package in R; Hothorn, Bretz, and Westfall 2008). To test whether P. rapae changed its oviposition behavior with increasing environmental complexity, we first evaluated how specialization was affected by nonhost complexity treatment using a GLM with Gaussian error structure. We also evaluated how landing preference was affected by nonhost complexity. When using host landing preference as a response variable (rather than as dependent variable as in earlier models), we calculated landing preference as the probability of landing on x number of cabbages in a sequence of 20 host landings. These models used binomial error structure with cabbage landings arbitrarily scored as “successes.” We were also interested in testing the consistency of landing preference in individual butterflies over time. To understand how landing preference changed over the course of a foraging bout, we divided a butterfly’s host landings into 2 periods: landings 1–10 and 11–20. We then modeled landing preference probability (probability of landing on x cabbages in 20 host landings) as a function of period and complexity treatment with individual as a random effect (“lme4” package in R; Bates et al. 2014). We also evaluated landing preference for cabbage or radish as a function of continuous landings within a trial using logistic regression, as follows. Each landing on a host (radish or cabbage) was assigned a number 1–20 indicating its position in the order of landings. For each landing, we scored cabbage as 1 and radish as 0 and modeled the probability of landing on cabbage as a function of landing number and plant community complexity with individual as a random effect. Total search time was modeled as a function of a butterfly’s accuracy, plant community complexity, and their interaction using a GLM with a Gamma distribution and log link. We evaluated the significance of terms with type-III likelihood-ratio tests (“car” package; Fox and Weisberg 2011). To test whether butterflies grew at a faster rate on cabbage or on radish, we modeled larval growth rate as a function of host type. RESULTS A total of 37 butterflies successfully completed foraging trials across three environments: complex (n = 16), simple-basil (n = 8), and simple-Swiss chard (n = 13). Across all three treatments, butterflies exhibited a variety of landing preferences and degrees of specialization, with some butterflies landing mostly on radish, some mostly on cabbage, and some visiting both equally (Figure 1). On average, butterflies visited 11.22 ± 0.64 (mean ± SE) cabbages in 20 host landings. A butterfly’s landing preference for cabbage, however, was not constant within an individual foraging bout; there was a higher probability that a butterfly would land on cabbage during its second ten landings than during its first 10 landings (β = 0.29 ± 0.13 SE, P = 0.03). On average, butterflies landed on 4.86 ± 0.36 (mean ± SE) cabbages during the first 10 landings and 6.35 ± 0.36 cabbages during the second 10 landings (Figure 2). Analysis using logistic regression yielded similar results; butterflies were more likely to land on cabbage as the number of host landings they made increased (β = 0.07 ± 0.01 SE, P < 0.01). Figure 1 View largeDownload slide Distribution of the number of landings on cabbage by individual butterflies during a single foraging bout of 20 host landings. A score of 0 implies that all 20 of a butterfly’s host landings were on radish; a score of 20 implies that all landings were on cabbage. On average, butterflies (n = 37) landed on 11.22 ± 0.11 (mean ± SE) cabbages in a foraging bout, indicated by the dashed line. Figure 1 View largeDownload slide Distribution of the number of landings on cabbage by individual butterflies during a single foraging bout of 20 host landings. A score of 0 implies that all 20 of a butterfly’s host landings were on radish; a score of 20 implies that all landings were on cabbage. On average, butterflies (n = 37) landed on 11.22 ± 0.11 (mean ± SE) cabbages in a foraging bout, indicated by the dashed line. Figure 2 View largeDownload slide Total cabbage landings by individual butterflies in the first period (Landings 1–10) and second period (Landings 11–20) of a foraging bout. Thin lines connect data points from the same individual butterfly and are ‘jittered’ for easier visualization. Thick black line and error bars denote the mean cabbage landings ±1.96 SE of all butterflies (n = 37) in the first and second periods. Overall, butterflies increased their landings on cabbage from the first period to the second period. Figure 2 View largeDownload slide Total cabbage landings by individual butterflies in the first period (Landings 1–10) and second period (Landings 11–20) of a foraging bout. Thin lines connect data points from the same individual butterfly and are ‘jittered’ for easier visualization. Thick black line and error bars denote the mean cabbage landings ±1.96 SE of all butterflies (n = 37) in the first and second periods. Overall, butterflies increased their landings on cabbage from the first period to the second period. Variation in the number of nonhost landings (accuracy) was best predicted by models containing host landing preference (number of cabbage landings; Table 1). Although the highest ranked model contained both host landing preference and nonhost complexity treatment (Akaike Weight = 0.51; Table 1) it was not distinguishable from a model containing only host landing preference (ΔAIC = 0.36; Table 1). Both of these models had Akaike Weights more than 5 points higher than that of the next best model and more than 7 points higher than the null model (Table 1). Analysis of the model with the lowest AICc value, containing both host landing preference and nonhost complexity, is reported in Table 2. Host landing preference was strongly associated with accuracy; butterflies that visited more cabbages also tended to visit fewer nonhosts (Table 2; Figure 3). Nonhost complexity was not a good predictor of accuracy (Table 2, Figure 4). Models of accuracy containing “landing preference” consistently outranked those containing “specialization,” indicating that a butterfly’s degree of specialization was a poor predictor of accuracy (Table 1). Table 1 Comparison of models predicting the accuracy (number of nonhost landings for every 20 host landings) of individual searching butterflies   Akaike Weights  deltaAIC  AICc  LL ratio  Nonhost Landings  Landing Pref. + Nonhost Complexity  0.51  0.00  185.57  −88.16  Landing Preference  0.43  0.36  185.93  −90.79  Landing Pref. × Nonhost Complexity  0.03  5.43  191.00  −88.10  ~ 1 NULL  0.01  7.31  192.88  −95.38  Nonhost Complexity  0.01  8.95  194.52  −93.90  Specialization  0.01  9.18  194.75  −95.20  Specialization + Nonhost Complexity  0.00  10.67  196.24  −93.49  Specialization × Nonhost Complexity  0.00  14.85  200.42  −92.81    Akaike Weights  deltaAIC  AICc  LL ratio  Nonhost Landings  Landing Pref. + Nonhost Complexity  0.51  0.00  185.57  −88.16  Landing Preference  0.43  0.36  185.93  −90.79  Landing Pref. × Nonhost Complexity  0.03  5.43  191.00  −88.10  ~ 1 NULL  0.01  7.31  192.88  −95.38  Nonhost Complexity  0.01  8.95  194.52  −93.90  Specialization  0.01  9.18  194.75  −95.20  Specialization + Nonhost Complexity  0.00  10.67  196.24  −93.49  Specialization × Nonhost Complexity  0.00  14.85  200.42  −92.81  A model with landing preference (total landings on cabbage) and nonhost complexity (diversity of nonhosts) was one of the best models predicting the accuracy of searching butterflies. View Large Table 2 Results of the best model predicting the accuracy (measured by number of nonhost landings for every 20 host landings) of individual searching butterflies   β  SE  z  P value  Nonhost Landings  Landing Preference   Total Number of Cabbage Landings  −0.04  0.01  −3.42  <0.001*  Nonhost Complexity   Simple Basil–Complex  −0.26  0.14  −1.86  0.15   Simple Swiss Chard–Complex  0.06  0.11  0.50  0.87   Simple Swiss Chard–Simple Basil  0.32  0.15  2.18  0.07    β  SE  z  P value  Nonhost Landings  Landing Preference   Total Number of Cabbage Landings  −0.04  0.01  −3.42  <0.001*  Nonhost Complexity   Simple Basil–Complex  −0.26  0.14  −1.86  0.15   Simple Swiss Chard–Complex  0.06  0.11  0.50  0.87   Simple Swiss Chard–Simple Basil  0.32  0.15  2.18  0.07  View Large Figure 3 View largeDownload slide The association between the total number of cabbage landings and the total number of nonhost landings made by individual butterflies. Each point represents data from the foraging trial of one individual butterfly, “jittered” for easier visualization. Line represents predicted values from the best-fit model. Butterflies that preferred cabbage over radish tended to land on fewer nonhosts (β = −0.04, P < 0.001). Figure 3 View largeDownload slide The association between the total number of cabbage landings and the total number of nonhost landings made by individual butterflies. Each point represents data from the foraging trial of one individual butterfly, “jittered” for easier visualization. Line represents predicted values from the best-fit model. Butterflies that preferred cabbage over radish tended to land on fewer nonhosts (β = −0.04, P < 0.001). Figure 4 View largeDownload slide The total number of nonhost landings made by individual butterflies while searching for host plants in simple (1 nonhost species) and complex (4 nonhost species) communities. Nonhost community complexity had no effect on butterfly accuracy. Bars represent means ± 1.96 SE. Figure 4 View largeDownload slide The total number of nonhost landings made by individual butterflies while searching for host plants in simple (1 nonhost species) and complex (4 nonhost species) communities. Nonhost community complexity had no effect on butterfly accuracy. Bars represent means ± 1.96 SE. Butterflies foraging in the complex nonhost environment did not specialize more than butterflies foraging in the simple basil environment (β = −0.06, P = 0.23) or butterflies foraging in the simple Swiss chard environment (β = −0.03, P = 0.55). Similarly, butterflies foraging in the complex nonhost environment had the same probability of landing on cabbages as butterflies foraging in the simple basil environment (β = −0.31, P = 0.11) and the simple Swiss chard environment (β = 0.07, P = 0. 69). A butterfly’s total time spent searching during a trial did not change with nonhost complexity (χ2 = 0.99, P = 0.61) and was not predicted by foraging accuracy (χ2 < 0.01, P = 0.99) or their interaction (χ2 = 2.52, P = 0.28). A total of 262 larvae were raised on cabbage plants (n = 130) and radish plants (n = 132). On average, larvae grew at a rate of 7.57 ± 0.12 mg/day (mean ± SE) when raised on cabbage plants and at a rate of 9.12 ± 0.12 mg/day when raised on radish plants. In our model, host identity had a significant effect on growth rate where larvae developed 1.54 ± 0.18 mg (mean ± SE) faster per day on radish than on cabbage (t = 8.77, P < 0.01). The reflectance spectra of host plants (cabbage and radish) and nonhost plants (basil, Swiss chard, lettuce, and pepper) were visually compared (Figure 5) from 300 to 700 nm, the spectral sensitivity range of the cabbage white butterfly (Shimohigashi and Tominaga 1991). The cabbage plants reflect light in the UV spectrum (300–400 nm) whereas the radish plants and other nonhost plants do not. Figure 5 View largeDownload slide The reflectance spectra of cabbage and radish host leaves (dark gray) compared to nonhost leaves (light gray). The reflectance spectrum of cabbage extends into the UV range (300–400 nm) whereas the spectra of radish and nonhost plants (pepper, Swiss chard, basil, lettuce) do not. Lines and ribbons represent the mean ± SE calculated from four individual plants of each species. Figure 5 View largeDownload slide The reflectance spectra of cabbage and radish host leaves (dark gray) compared to nonhost leaves (light gray). The reflectance spectrum of cabbage extends into the UV range (300–400 nm) whereas the spectra of radish and nonhost plants (pepper, Swiss chard, basil, lettuce) do not. Lines and ribbons represent the mean ± SE calculated from four individual plants of each species. DISCUSSION Selective information processing as a mechanism of individual specialization If selective information processing drives individual specialization, we expected the more specialized individuals to be more accurate than comparatively generalized individuals. In our study, however, only certain individuals that visited one resource were more accurate than those that visited two. Only the butterflies that primarily visited cabbage landed on fewer nonhosts than butterflies that landed on a mix of both hosts; butterflies that primarily visited radish displayed no such foraging advantage. Furthermore, despite variation in individual landing preference for cabbage vs. radish, all butterflies increased their landing preference for cabbage over time. These results offer mixed support for the hypothesis that information-processing limitations underlie patterns of individual specialization. Specialization was not associated with foraging advantages for all the resources in our assay perhaps because visiting one resource may only be advantageous if it is cues are easy to distinguish from nonresources or from the background environment. In many systems, when background contrast negatively affects the detectability of one of the target resources, another target becomes preferred (Forrest and Thomson 2009; Muchhala and Serrano 2015). In our system, cabbage may be the easier of the 2 “targets” (radish, cabbage) to distinguish from the many “distractors” (swiss chard, pepper, basil, lettuce). Cabbage leaves are covered in a wax that reflects UV wavelengths, making them visually distinct from radish and from the distractor nonhost plants (Figure 5). Cabbage white butterflies possess photoreceptors that are sensitive to these wavelengths (Shimohigashi and Tominaga 1991) and rely heavily on vision to process their environment (Snell-Rood et al. 2009) and make landing decisions (Tsuji and Coe 2014). We quantified the distinctiveness of cabbage compared to other plants in terms of vision, but it is also possible that cabbage is distinct in a completely different sensory modality given that butterflies use a combination of vision, olfaction, and chemosensation to locate hosts (Hern et al. 1996). Alternatively, the cues associated with cabbage might be attractive to butterflies not because the cues are distinguishable, but because of a tighter co-evolutionary relationship with the genus Brassica (cabbage) than with the genus Raphanus (radish), resulting in a sensory or learning bias associated with Brassica cues. The evolution of such a sensory or learning bias would be particularly expected if female landing preferences were correlated with larval performance (Gripenberg et al. 2010). However, in our greenhouse system, female cabbage whites, on average, preferred to oviposit on cabbage plants even though their larvae grow faster on radish plants. The observation that females prefer to land on a plant that is apparently less suitable for larval growth suggests the possibility of a tradeoff between larval growth rate on a host and other ecological factors related to host plant choice (presence of natural enemies, competition) or factors related to adult foraging choices (host relative abundance, host detectability/distinguishability). Alternatively, cabbage white preference for Brassica (cabbage) over Raphanus (radish) might reflect a past selection bias, if, for example, Brassica were more nutritious than Raphanus prior to their cultivation for agricultural purposes. Adult female preference for the lower quality host in our system may also be a product of temporal or spatial fluctuations in host quality in the field that we have not considered (Cronin et al. 2001) or because the cues associated with cabbage are indicators of quality in another context (e.g. nectar foraging, mate choice). Although our study suggests that information processing and specialization are related in certain cases, this study does not provide evidence that selective information processing promotes different specializations among different individuals. In the individual specialization literature, researchers are interested in the idea that individuals specialize on different resources from each other, enhancing the heterogeneity of the community (Bolnick et al. 2003; Bolnick et al. 2011). In our study, although we see variation in host landing preference among individuals during the first 10 host landings, all butterflies increased their landing preference for cabbage during the second ten host landings (Figure 2). If all individuals are becoming specialists on the same resource over time, a more homogenous community will emerge rather than the heterogeneous community that the individual specialization literature anticipates. Limits to information processing may indeed promote specialization but might not have the effect of diversifying resource use within a population, particularly if specializations are influenced by sensory or learning biases. Different ecological and evolutionary dynamics would result if specializing led to homogenous resource use rather than heterogeneous resource use among individuals (Bolnick et al. 2003). Implications of information processing in complex environments In this study, the complex environment did not appear to give individuals that visited one resource an advantage over individuals that visited two resources nor did the complex environment cause butterflies to be more specialized than butterflies in the simple environment. These results suggest several possibilities: 1) our complexity manipulation was imperceptible to the butterflies, 2) foraging primarily on one resource did not alleviate the cognitive challenges of complex nonhost environments, or 3) butterflies compensated for the difficulty of complex environments by changing the speed with which they made decisions. Unfortunately, the limitations of our experimental design do not allow us to definitively distinguish among these 3 scenarios. Although it is tempting to conclude that nonhost complexity had no effect on foraging performance in our experiment, our design has limitations that make negative results difficult to interpret. We expected that increasing the diversity of nonhost plants would increase the diversity of leaf colors, odors, shapes, and chemistries in the environment, thereby increasing the overall stimuli that a butterfly would need to process while searching. Although butterflies do use all of these cues to recognize host plants (color: Traynier 1984, Snell-Rood and Papaj 2009; odor: Ikeura, Kobayashi, and Hayata 2010; shape: Papaj 1986), we cannot explicitly confirm that the cabbage whites perceived each of the nonhosts (basil, Swiss chard, lettuce, pepper) as different from each other. Furthermore, given that the challenges of a generalist lie in keeping track of multiple types of host plants (rather than multiple types of nonhost plants), it is also possible that our nonhost manipulation had no effect on an individual’s tendency to specialize because specialization is most strongly driven by the relative abundance and diversity of host plants in the environment. Secondly, we used only 1 type of complex array (with one set of nonhost species) and only 2 types of simple arrays in our experiment. Using a broader range of both complex and simple arrays might help us see a clearer effect of complexity on search behavior. Our observation that butterflies searching in one of the simple arrays (basil) landed on marginally fewer nonhosts than butterflies searching in the other simple array (Swiss chard) suggests that the identity of the nonhost community does have the potential to influence foraging performance. Perhaps, over a wider range of simple and complex nonhost compositions involving many different nonhost species, we might have found an effect of complexity that was not observed for the limited range of nonhost compositions used. Butterflies might have compensated for the difficulty of the complex environment by taking a longer time to make decisions, as predicted by a speed-accuracy tradeoff. This tradeoff is often more prominent in difficult decision-making environments (Dyer and Chittka 2004). However, the total search time required for a butterfly to make 20 host landings was similar across all three foraging environments and unrelated to a butterfly’s accuracy. Our measure of search time, though, should be taken with some caution. Our measure included not only a butterfly’s flight time between resources, but also the time required to lay an egg, which varies greatly among individuals and is not directly related to a pre-landing decision time. Furthermore, because butterflies are ectotherms, brief cloud cover or temperature shifts during a trial can have effects on foraging speed that are independent of the difficulty of the search task. Thus, butterflies may still be compensating for complexity by changing decision speed in ways that the observer could not detect. Although butterflies might not have been able to perceive our complexity manipulation, there is evidence from the literature to suggest that the diversity of distractors in a search task may not influence foraging efficiency even when searchers can perceive the change in complexity. For example, bees foraging for targets among only blue distractors had the same error rate and decision time as bees foraging for the same targets among mixed-color distractors (Spaethe et al. 2006). Kostenko et al. (2015) found that the diversity of nonhost plants in a field experiment did not influence a parasitoid’s host location success but that structural complexity (no vegetation, mowed vegetation, tall vegetation) did influence aggregation behavior. In the agricultural literature, planting a diversity of companion plants with a target crop often has variable results on pest populations (Andow 1991; for Brassicaeceae crops specifically, Hooks and Johnson 2003). It is difficult to draw direct conclusions about the effect of nonhost plant diversity on the insect search process from many field studies because manipulating companion crop diversity alters other confounding factors like floral resources, natural enemies, and belowground plant competition (Letourneau et al. 2011). Our study, which controlled for all of these factors, suggests that, in some cases, the diversity of nonhosts may not hamper an insect’s ability to locate hosts. There may be other features of the distractor community that have a greater effect on decision-making than diversity. For a sequentially foraging animal like the cabbage white butterfly, decision-making in a series is more likely to be influenced by the total number of times the searcher encounters distractors in a sequence (influenced by the ratio of distractors to targets; Treisman and Gelade 1980) rather than the different kinds of distractors (influenced by diversity). When bees were trained to distinguish between rewarding target flowers and nonrewarding distractors, error rate increased 4-fold and decision time increased 1.5-fold as the number of distractor flowers in the assay increased from one to fifteen (Spaethe, Tautz, and Chittka 2006). The identity of distractors may also influence the difficulty of a foraging task. When a series of 24 distractor nonhost plants were intercropped (each individually) with a target host Brassica plant, the number of eggs laid by the cabbage root fly was affected differently by each type of distractor (Finch et al. 2003). A variety of mechanisms might make certain nonhost plants more effective at masking host plants than others. For example, searching might be more difficult when distractor properties make discriminating between targets and distractors more difficult. The error rate of foraging bees increased when target and distractor flowers had similar colors (Dyer and Chittka 2004). This idea that distractor identity influences foraging accuracy is partially supported in our data. Butterflies foraging in the basil nonhost environment were marginally (though not significantly) more accurate than butterflies foraging in our Swiss chard nonhost environment (Table 2; Figure 4). Considering that basil has a distinct olfactory profile, it might have been easier for butterflies to discriminate between cabbage/radish targets and very obvious basil distractors than between cabbage/radish targets and Swiss chard distractors. These observations suggest nonhost identity might create cognitive challenges for the butterflies, possibly resulting in individual specialization. CONCLUSIONS This work provides partial support for the idea that selective attention drives individual specialization because certain individuals that visited 1 resource (cabbage) foraged more accurately than individuals that visited 2 resources (cabbage and radish). However, because not all resources provided the foraging benefits associated with specialization, selective attention may not drive individuals to develop specializations that are different from each other. If we want to further understand how behavioral mechanisms (like learning or information processing) influence individual specialization on various timescales, experiments that manipulate both a forager’s experience and environment over time will be particularly important. Such detailed individual-level data, although difficult to obtain, are necessary to understand how specialized resource preferences develop, how long they last, and how they interact with environmental context. Further clarifying the mechanisms behind individual specialization will allow us to make concrete predictions about the environmental factors that drive specialization and the degree of resource use heterogeneity among individuals in a population. FUNDING This work was supported by the Bell Museum of Natural History (Dayton Bell Museum Fund Fellowship). The Snell-Rood lab was supported in part through NSF-IOS-1354737 during this time. Data accessibility Analyses reported in this article can be reproduced using the data provided by Steck and Snell-Rood (2017). Many thanks to Kush Patel, Monika Malek, Rhea Smykalski, Chase Marx, and Regina Kurandina for help collecting butterflies, caring for them, rearing larvae, and growing plants. We appreciated discussion and feedback from the Snell-Rood lab and from post-docs Eli Swanson and Eric Lind. We also thank three anonymous reviewers and the handling editor for very thoughtful, constructive comments that greatly improved the quality of this manuscript. REFERENCES Andow D. 1991. Vegetational diversity and arthropod population response. Annu Rev Entomol . 36: 561– 586. Google Scholar CrossRef Search ADS   Araújo MS, Bolnick DI, Layman CA. 2011. The ecological causes of individual specialisation. Ecol Lett . 14: 948– 58. Google Scholar CrossRef Search ADS PubMed  Bates D, Maechler M, Bolker B, Walker S. 2014. lme4: Linear mixed-effects models using Eigen and S4. R package version1.1–7. Available from: http://CRAN.Rproject.org/package=lme4. Bauerfiend SS, Fischer K. 2008. Maternal body size as a morphological constraint on egg size and fecundity in butterflies. Bas Appl Ecol . 9: 443– 451. Google Scholar CrossRef Search ADS   Bernays EA, Wcislo WT. 1994. Sensory capabilities information processing and resource specialization. Q Rev Biol . 69: 187– 204. Google Scholar CrossRef Search ADS   Bernays EA. 1997. Feeding by lepidopteran larvae is dangerous. Ecol Entomol . 22: 121– 123. Google Scholar CrossRef Search ADS   Bernays EA. 1998. The value of being a resource specialist: behavioral support for a neural hypothesis. Am Nat . 151: 451– 464. Google Scholar CrossRef Search ADS PubMed  Bernays EA. 2001. Neural limitations in phytophagous insects : implications for diet breadth and evolution of host affiliation. Annu Rev Entomol . 46: 703– 727. Google Scholar CrossRef Search ADS PubMed  Bernays EA, Singer MS, Rodrigues D. 2004. Foraging in nature: Foraging efficiency and attentiveness in caterpillars with different diet breadths. Ecol Entomol . 29: 389– 397. Google Scholar CrossRef Search ADS   Bolnick DI, Yang LH, Fordyce JA, Davis JM, Svanbäck R. 2002. Measuring individual-level resource specialization. Ecology . 83: 2936– 2941. Google Scholar CrossRef Search ADS   Bolnick DI, Svanbäck R, Fordyce JA, Yang LH, Davis JM, Hulsey CD, Forister ML. 2003. The ecology of individuals: incidence and implications of individual specialization. Am Nat . 161: 1– 28. Google Scholar CrossRef Search ADS PubMed  Bolnick DI, Amarasekare P, Araújo MS, Bürger R, Levine JM, Novak M, Rudolf VHW, Schrieber SJ, Urban MC, Vasseur DA. 2011. Why intraspecific trait variation matters in community ecology. Trends Ecol Evol . 26: 183– 192. Google Scholar CrossRef Search ADS PubMed  Brosi BJ. 2016. Pollinator specialization: from the individual to the community. New Phytol . 210: 1190– 1194. Google Scholar CrossRef Search ADS PubMed  Chittka L, Skorupski P, Raine NE. 2009. Speed-accuracy tradeoffs in animal decision making. Trends Ecol Evol . 34: 400– 407. Google Scholar CrossRef Search ADS   Cronin JT, Abrahamson WG, Craig TP. 2001. Temporal variation in herbivore host-plant preference and performance: constraints on host-plant adaptation. Oikos . 93: 312– 320. Google Scholar CrossRef Search ADS   Dall SRX, Cuthill IC. 1997. The information costs of generalism. Oikos . 80: 197– 202. Google Scholar CrossRef Search ADS   Dall SRX, Bell AM, Bolnick DI, Ratnieks FLW. 2012. An evolutionary ecology of individual differences. Ecol Lett . 15: 1189– 1198. Google Scholar CrossRef Search ADS PubMed  Darimont CT, Paquet PC, Reimchen TE. 2007. Stable isotopic niche predicts fitness of prey in a wolf-deer system. Biol J Linn Soc . 90: 125– 137. Google Scholar CrossRef Search ADS   Dieckmann U, Doebeli M. 1999. On the origin of species by sympatric speciation. Nature . 400: 354– 357. Google Scholar CrossRef Search ADS PubMed  Duffy MA. 2010. Ecological consequences of intraspecific variation in lake Daphnia. Freshwater Biol . 55: 995– 1004. Google Scholar CrossRef Search ADS   Dukas R, Kamil AC. 2001. Limited attention : the constraint underlying search image. Behav Ecol . 12: 192– 199. Google Scholar CrossRef Search ADS   Dukas R. 2002. Behavioural and ecological consequences of limited attention. Philos Trans R Soc Lond B Biol Sci . 357: 1539– 1547. Google Scholar CrossRef Search ADS PubMed  Dukas R. 2004. Causes and consequences of limited attention. Brain Behav Evol . 63: 197– 210. Google Scholar CrossRef Search ADS PubMed  Dusenbery DB. 1992. Sensory ecology: how organisms acquire and respond to information . New York (NY): W.H. Freeman. Dyer AG, Chittka L. 2004. Biological significance of distinguishing between similar colours in spectrally variable illumination: Bumblebees (Bombus terrestris) as a case study. J. Comp. Physiol. A Neuroethol. Sensory, Neural, Behav. Physiol . 190: 105– 114. Google Scholar CrossRef Search ADS   Egan SP, Funk DJ. 2006. Individual advantages to ecological specialization: insights on cognitive constraints from three conspecific taxa. Proc Biol Sci . 273: 843– 848. Google Scholar CrossRef Search ADS PubMed  Estes JA, Riedman ML, Staedler MM, Tinker MT, Lyon BE. 2003. Individual variation in prey selection by sea otters: patterns, causes and implications. J Anim Ecol . 72: 144– 155. Google Scholar CrossRef Search ADS   Finch S, Billiald H, Collier RH. 2003. Companion planting - do aromatic plants disrupt host-plant finding by the cabbage root fly and the onion fly more effectively than non-aromatic plants? Entomol Exp Appl . 109: 183– 195. Google Scholar CrossRef Search ADS   Forrest J, Thomson JD. 2009. Background complexity affects colour preference in bumblebees. Naturwissenschaften  96: 921– 925. Google Scholar CrossRef Search ADS PubMed  Fox LR, Morrow PA. 1981. Specialization: species property or local phenomenon? Science . 211: 887– 893. Google Scholar CrossRef Search ADS PubMed  Fox J, Weisberg S. 2011. An {R} companion to applied regression , Second Edition. Thousand Oaks (CA): Sage. Available from: http://socserv.socsci.mcmaster.ca/jfox/Books/Companion. García-Barros E. 2000. Body size, egg size, and their interspecific relationships with ecological and life history traits in butterflies (Lepidoptera: Papilionoidea, Hesperiodiea). Biol J Linn Soc . 70: 251– 284. Google Scholar CrossRef Search ADS   Gols R, Bukovinszky T, Hemerik L, Harvey JA, Van Lenteren JC, Vet LEM. 2005. Reduced foraging efficiency of a parasitoid under habitat complexity: implications for population stability and species coexistence. J Anim Ecol . 74: 1059– 1068. Google Scholar CrossRef Search ADS   Gripenberg S, Mayhew P, Parnell M, Roslin, T. 2010. A meta-analysis of preference-performance relatinoships in phytophagous insects. Ecol Lett . 13: 383– 393. Google Scholar CrossRef Search ADS PubMed  Heinrich B. 1976. The foraging specializations of individual bumblebees. Ecol Mono . 46: 105– 128. Google Scholar CrossRef Search ADS   Heinrich B. 1993. Caterpillars: ecological and evolutionary constraints on foraging . New York (NY): Chapman & Hall. Hern A, Edwards-Jones G, McKinlay RG. 1996. A review of the pre-oviposition behaviour of the small cabbage white butterfly, Pieris rapae (Lepidoptera : Pieridae). Annu. Appl Biol  128: 349– 371. Google Scholar CrossRef Search ADS   Hooks CRR, Johnson MW. 2003. Impact of agricultural diversification on the insect community of cruciferous crops. Crop Prot . 22: 223– 238. Google Scholar CrossRef Search ADS   Hothorn T, Bretz F, Westfall P. 2008. Simultaneous inference in general parametric models. Biom J . 50: 346– 363. Google Scholar CrossRef Search ADS PubMed  Hughes AR, Inouye BD, Johnson MTJ, Underwood N, Vellend M. 2008. Ecological consequences of genetic diversity. Ecol Lett . 11: 609– 623. Google Scholar CrossRef Search ADS PubMed  Ikeura H, Kobayashi F, Hayata Y. 2010. How do Pieris rapae search for Brassicaceae host plants? Biochem Sys Ecol . 38: 1199– 1203. Google Scholar CrossRef Search ADS   Jaumann S, Snell-Rood EC. 2017. Trade-offs between fecundity and choosiness in ovipositing butterflies. Anim Behav . 123: 433– 440. Google Scholar CrossRef Search ADS   Janz N, Nylin S. 1997. The role of female search behaviour in determining host plant range in plant feeding insects: a test of the information processing hypothesis. Proc R Soc B Biol Sci . 264: 701– 707. Google Scholar CrossRef Search ADS   Janz N. 2003. The cost of polyphagy: oviposition decision time vs error rate in a butterfly. Oikos . 100: 493– 496. Google Scholar CrossRef Search ADS   Johnson CK, Tinker MT, Estes JA, Conrad PA, Staedler M, Miller MA, Jessup DA, Mazet JAK. 2009. Prey choice and habitat use drive sea otter pathogen exposure in a resource-limited coastal system. Proc Natl Acad Sci USA . 106: 2242– 2247. Google Scholar CrossRef Search ADS PubMed  Kastner S, Ungerleider LG. 2000. Mechanisms of visual attention in the human cortex. Annu Rev Neurosci . 23: 315– 341. Google Scholar CrossRef Search ADS PubMed  Kocher SD, Williams EH. 2000. The diversity and abundance of North American butterflies cary with habitat disturbance and geography. J Biogeogr . 27: 785– 794. Google Scholar CrossRef Search ADS   Kostenko O, Lammers M, Grootemaat S, Kroon T, Harvey JA, Van Geem M, Bezemer TM. 2015. Effects of plant diersity and structural complexity on parasitoid behavior in a field experiment. Ecol Entomol . 40: 748– 758. Google Scholar CrossRef Search ADS   Layman CA, Newsome SD, Gancos Crawford T, Crawford TG. 2015. Individual-level niche specialization within populations: emerging areas of study. Oecologia . 178: 1– 4. Google Scholar CrossRef Search ADS PubMed  Letourneau DK, Armbrecht I, Rivera BS, Lerma M, Carmona EJ, Daza MC, Escobar S, Galindo V, Gutierrez C, Lopez SDet al.   2011. Does plant diversity benefit agroecosystems? Ecol Appl . 21: 9– 21. Google Scholar CrossRef Search ADS PubMed  Lomnicki A. 1978. Individual differences between animals and the natural regulation of their numbers. J Anim Ecol . 47: 461– 475. Google Scholar CrossRef Search ADS   Maia R, Eliason CM, Bitton PP, Doucet SM, Shawkey MD. 2013. pavo: an R Package for the analysis, visualization and organization of spectral data. Methods Ecol Evol . 4: 906– 913. Mazerolle MJ. 2015. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 2.0–3. Available from: http://CRAN.R-project.org/package=AICcmodavg. Moran J, Desimone R. 1985. Selective attention gates visual processing in the extrastriate cortex. Science . 229: 782– 784. Google Scholar CrossRef Search ADS PubMed  Muchhala N, Serrano D. 2015. The complexity of background clutter affects nectar bat use of flower odor and shape cues. PLoS One  10: 1– 12. Google Scholar CrossRef Search ADS   Papaj D. 1986. Conditioning of leaf-shape discrimination by chemical cues in the butterfly, Battus philenor. Anim Behav . 34: 1281– 1288. Google Scholar CrossRef Search ADS   Perfecto I, Vet LEM. 2003. Effect of a non-host plant on the location behavior of two parasitoids: the tritrophic system of Cotesia spp. (Hymenoptera : Braconidae), Pieris rapae (Lepidoptera : Pieridae), and Brassica oleraceae. Environ Entomol . 32: 163– 174. Google Scholar CrossRef Search ADS   R Core Team. 2013. R: A language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Available from: http://www.R-project.org/. Raitanen J, Forsman JT, Kivelä SM, Mäenpää MI, Välimäki P. 2014. Attraction to conspecific eggs may guide oviposition site selection in a solitary insect. Behav Ecol . 25: 110– 116. Google Scholar CrossRef Search ADS   Reimchen TE, Nosil P. 2001. Dietary differences between phenotypes with symmetrical and asymmetrical pelivs in the stickleback Gasterosteus aculeatus. Can J Zool . 79: 533– 539. Google Scholar CrossRef Search ADS   Renwick J, Chew F. 1994. Oviposition behavior in lepidoptera. Annu Rev Entomol . 39: 377– 400. Google Scholar CrossRef Search ADS   Scott JA. 1986. The Butterflies of North America: A Natural History and Field Guide . Standford, California: Stanford University Press. Shimohigashi M, Tominaga Y. 1991. Identification of UV, green and red receptors, and their projection to lamina in the cabbage butterfly, Pieris rapae. Cell Tissue Res . 263: 49– 59. Google Scholar CrossRef Search ADS   Snell-Rood EC, Papaj DR. 2009. Patterns of phenotypic plasticity in common and rare environments: a study of host use and color learning in the cabbage white butterfly Pieris rapae. Am Nat . 173: 615– 631. Google Scholar CrossRef Search ADS PubMed  Snell-Rood EC, Papaj DR, Gronenberg W. 2009. Brain Size: A global or induced cost of learning? Brain Behav Evol . 73: 111– 128. Google Scholar CrossRef Search ADS PubMed  Spaethe J, Tautz J, Chittka L. 2006. Do honeybees detect colour targets using serial or parallel visual search? J Exp Biol . 209: 987– 993. Google Scholar CrossRef Search ADS PubMed  Steck MK, Snell-Rood E. 2017. Data from: specialization and accuracy of host-searching butterflies in complex and simple environments. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.s70m4. Svanbäck R, Bolnick DI. 2007. Intraspecific competition drives increased resource use diversity within a natural population. Proc Biol Sci . 274: 839– 844. Google Scholar CrossRef Search ADS PubMed  Terraube J, Arroyo B, Madders M, Mougeot F. 2011. Diet specialisation and foraging efficiency under fluctuating vole abundance: A comparison between generalist and specialist avian predators. Oikos  120: 234– 244. Google Scholar CrossRef Search ADS   Traynier RMM. 1984. Associative learning in the ovipositional behavior of the cabbage white butterfly, Pieris rapae. Phys Entomol . 9: 465– 472. Google Scholar CrossRef Search ADS   Treisman AM, Gelade G. 1980. A feature-integration theory of attention. Cogn Psychol . 12: 97– 136. Google Scholar CrossRef Search ADS PubMed  Tsuji J, Coe L. 2014. Effects of foliage color on the landing response of Pieris rapae (Lepidoptera: Pieridae). Environ Entomol . 43: 989– 994. Google Scholar CrossRef Search ADS PubMed  Van Valen L. 1965. Morphological variation and width of ecological niche. Am Nat . 99: 377– 390. Google Scholar CrossRef Search ADS   Wee B, Singer M. 2007. Variation among individual butterflies along a generalist – specialist axis : no support for the “ neural constraint “ hypothesis. Ecol Entomol . 32: 257– 261. Google Scholar CrossRef Search ADS   Xiao F, Cuthill IC. 2016. Background complexity and the detectability of camouflaged targets by birds and humans. Proc R Soc B . 283: 20161527. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Behavioral EcologyOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial