Arthropod Demography, Distribution, and Dispersion in a Novel Trap-Cropped Cotton Agroecosystem

Abstract

Vernonia [Vernonia galamensis (Cass.) Less.] (Asterales: Asteraceae) was examined as a potential trap crop for the cotton (Gossypium hirsutum L., Malvales: Malvaceae) arthropod complex. Four rows of vernonia were embedded within a 96-row cotton field. The abundance of true bug pests, true bug predators, and spiders were determined by whole-plant and sweep net sampling procedures during the early, middle, and late phases of the cotton-growing season. The census data showed that the arthropods had a strong preference for the vernonia trap crop throughout the cotton-growing season. The movement of the arthropods from the trap crop into cotton was also measured using the protein immunomarking technique as a mark–capture procedure. The arthropods inhabiting the vernonia trap crop were marked directly in the field with a broadcast spray application of egg albumin (protein) during each phase of the study. In turn, the captured specimens were examined for the presence of the mark by an egg albumin-specific enzyme-linked immunosorbent assay. Very few marked specimens were captured beyond the vernonia trap crop 1, 3, and 6 d after each marking event. The arthropods’ strong attraction and fidelity to vernonia indicate that it could serve as a trap crop for cotton pests and a refuge for natural enemies.

The western tarnished plant bug, Lygus hesperus Knight (Hemiptera: Miridae), is a major pest of cotton in the United States (Strong 1970). Lygus has a wide host range that is well documented (Scott 1977). Although cotton is a suitable host plant for lygus, it has a strong preference for other plant species. For example, lygus prefer alfalfa over cotton, and strips of alfalfa ‘trap crops’ planted adjacent to cotton can reduce damage to cotton squares (Stern et al. 1964). However, alfalfa only provides an ephemeral habitat for lygus because it is frequently harvested throughout the cotton-growing season (Goodell 2009). When the alfalfa is cut, the lygus disperse into the cotton to feed and reproduce (Godfrey and Leigh 1994, Sivakoff et al. 2012). Since lygus feed and reproduce on cotton but are more attracted to a wide variety of other plant species, it seems plausible that a summer-stable plant could serve as a trap crop for this pest and other potential cotton pests. Additionally, flower strips have been shown to enhance ecosystem services for adjacent crops by providing a habitat for natural enemies and pollinators (Albrecht et al. 2020). However, opposing push–pull dynamics, i.e., the tendency of the flower strip to draw in arthropods versus its spillover into the crop, means this method is not effective in every case. The efficacy of the flower strip will depend on how well it acts as a sink for pests while generating spillover of natural enemies and pollinators (Tscharntke et al. 2016, Karp et al. 2018).

Vernonia [Vernonia galamensis (Cass.) Less; Asteraceae] is an annual plant that is native to eastern Africa that could serve as a trap crop or flower strip for cotton pests, natural enemies, and pollinators in the United States. In the current literature, this plant species is also referred to as Centrapalus pauciflorus (Willd.) H. Rob. (Robinson 1999, Todd et al. 2018). Itis a desert-adapted and low agronomic input crop that is planted in the spring and grows and blooms alongside cotton during the harsh summer months. Also, like cotton, vernonia can be cultivated as a row crop. Little is known about the arthropod complex inhabiting vernonia, but a feeding choice test showed that lygus is highly attracted to blooming vernonia (Hagler et al. 2016).

We hypothesized that the strategic placement of vernonia plants within a cotton field could serve as a preferable and stable habitat for lygus and other potential cotton pests. The vernonia trap crop could also provide a summer refuge for predators, parasitoids, and pollinators. This study was designed to characterize and quantify arthropod population dynamics and dispersal behavior of arthropods inhabiting a cotton field that was interplanted with vernonia. We used both an absolute whole-plant sampling scheme and a relative sweep net sampling scheme to characterize the demography and distribution of the arthropod complex at the study site. We also studied the retention potential of vernonia using the protein immunomarking technique (PIT) to monitor arthropod migration from the trap crop into the cotton throughout the cotton-growing season.

Our ultimate goals are to 1) reduce the dependency on broad-spectrum pesticides in cotton (which would reduce the occurrence of pesticide resistance), 2) increase biological control ecosystem services through greater conservation of natural enemies, and 3) increase pollinator ecosystem services by providing a more diverse and favorable habitat for pollinators during the summer months.

Materials and Methods

Propagation of Vernonia

Vernonia germplasm (seed) used in this study is described by Dierig et al. (2006). The seeds were obtained from seed stock maintained at the Arid-Land Agricultural Research Center, Maricopa, AZ. Individual seeds were sown in 6.3 cm² × 6.3 cm deep pots containing a standard soil mixture on 31 January and 16 February 2018. The seeded pots were maintained in a greenhouse set at 18°C (night) to 30°C (day) with a relative humidity of 30%. The plants were watered as needed. A 1:1 mixture of all-purpose Scotts Miracle-Gro Excel (21-5-20) and Cal-mag Miracle-Gro Professional (15-5-15) was applied (50 ml/plant) at a rate of 1% shortly after the seeded plants emerged and then reapplied approximately 1 mo later. The plants, ranging in height from 15 to 30 cm, were transplanted at the study site described below.

Study Site

The study was conducted in a 0.87-ha upland cotton field (Gossypium hirsutum L.) located at the University of Arizona Maricopa Agricultural Center (MAC), Maricopa, AZ (33.068°N, 11.971°W). Vernonia plants were transplanted at 1-m intervals at the study site on 13 April 2018. The cottonseed was planted on 20 April 2018. The cotton and vernonia plants were grown throughout the study period using standard agronomic practices used at the MAC farm for cotton production.

The experiment consisted of an early-, middle-, and late-season cotton growth phase. The data collected each growth phase were considered as independent studies. The various stages of the study were initiated on 26 June, 24 July, and 21 August 2018, respectively. A diagram of the field site is shown in Fig. 1. The field contained eight spans, each 12.2 m wide, of the crop. Each span consisted of 12 rows that were 118.5 m in length with a row spacing of 1.02 m. The spans were further divided into eight subplots that were 12.2 m long. Two fallow rows separated each span. The experiment was divided into three main treatment sampling zones. The two spans of the crop on the eastern side of the field were designated as the vernonia trap crop (VTC) treatment zone. Each of these spans contained 10 rows of cotton (rows 3–10) in between two rows (rows 1 and 12) of vernonia. The two spans on the western side of the field were designated as the cotton control (COT) treatment zone. The COT zone, which was the farthest possible from the VTC zone, consisted of 12 rows of cotton. The VTC and COT zones represent the primary focus of our sweep net sampling effort described below. The four spans (48 rows total) of the crop between the VTC and COT zones only contained cotton. They served as a buffer (BUF) treatment zone for the main VTC and COT treatments. The VTC and the COT main treatments were subdivided into two additional sampling zones. Rows 1 and 12 in the VTC and COT spans were designated as edge (E) sampling zones, and the 10 rows between them were designated at middle (M) sampling zones. As such, the four VTC-E rows contained the vernonia trap crop. The four COT-E rows were the complimentary cotton control rows to the vernonia trap crop. The VTC-E and COT-E zones represent the focus of the whole-plant sampling effort described below. The VTC-M and COT-M zones were also chosen as intense sweep net sampling areas to provide a better account (i.e., increase the sample size) for the short-range movement of arthropods from the vernonia.

Fig. 1.

Diagram of the 0.87-ha study site showing the sample sites. The abbreviation used for each zone designation is defined in the text. The black rectangles depict the location of the 100 plots where sweep net samples (25 sweeps per sample) were taken 1, 3, and 6 d after each marking event (n = 300 per growth phase). The red rectangles depict the 16 plots where the whole-plant samples were taken 1, 3, and 6 d after each marking event (n = 48 per growth phase).

Arthropod Mark–Capture Procedure

The resident arthropods inhabiting the four rows of vernonia (VTC-E) were marked on the dates given above with egg albumin contained in commercial chicken egg whites (All Whites 100% Liquid Egg Whites; Papetti Foods, Elizabeth, NJ). The egg whites were diluted to 10% with tap water. The 10% protein solution was applied with a gas-powered backpack sprayer (#MD155DX Mist Duster; Maruyama, Denton, TX). The mark was topically applied to each 118.5 m VTC-E row at a rate of 1 l per 6 m of row. Arthropods were collected after marking using the whole-plant and sweep net sampling protocols described below.

Whole-Plant Sampling Procedure

Due to labor intensity, whole-plant samples were only taken in the COT-E and VTC-E sampling zones (red rectangles shown in Fig. 1) and only during the early and middle phases of the cotton season. The whole-plant sampling protocol is described in detail by Hagler et al. (2014). Briefly, sleeve cages made from fine mesh organdy fabric were placed at the base of three vernonia and three cotton plants in each of the plots located in the VTC-E and COT-E sampling zones (Fig. 1). As such, eight whole-plant samples were taken from the VTC and COT main treatment zones on each of the three sampling dates (n = 24 samples per plant type). The whole-plant cages were placed in the field the day before each protein-marking event. The bottom of each sleeve cage was tied at the base of each plant with a nylon zip tie. The following day, 2 hr after the protein mark was sprayed on the vernonia, the top of each sleeve cage was pulled up and over each plant, and the top was sealed with a zip tie. The native arthropods entrapped in each cage roamed freely within the enclosure for either 1, 3, or 6 d after marking. At the end of each postmarking sampling interval, one of the three caged plants from each subplot was cut just below the bottom zip tie, and the entire caged plant was placed into a labeled plastic trash bag. Each plant was frozen within 30 min at −60°C. The whole-plant samples were processed by removing them from the freezer and carefully searching for the arthropods enclosed in each cage. The arthropods were identified to genus or species and tallied to determine per-capita estimate of the various taxa inhabiting an individual vernonia and cotton plant located in the VTC-E and COT-E treatments, respectively.

The capture rates of many of the taxa were too low for meaningful statistical analysis (e.g., too many zero counts per sample unit). As such, we pooled the taxa into three broad categories. The categories included 1) true bug pests, 2) true bug predators, and 3) spiders. The true bug pests consisted of known herbivores (albeit, not all of which are major cotton pests) and omnivores that are known cotton pests (e.g., lygus bugs). The true bug predators consisted of known carnivores (e.g., assassin bugs) and omnivores that are known predators (e.g., big-eyed bugs, minute pirate bugs, etc.; Hagler and Naranjo 1994, Hagler and Blackmer 2013). The spiders included all the spider taxa we encountered. Differences in arthropod abundance of each taxon between the VTC-E and COT-E treatments were analyzed by the Student’s t-test (JMP, Version 14.2 2019). However, the count data did not meet the assumptions needed for a valid t-test. Therefore, significant differences in abundance were determined using the nonparametric Mann–Whitney two-sample rank-sum test. All results are presented as the mean (±SD) number of arthropods collected per plant sample.

The arthropods collected in the whole-plant samples were also examined for the presence of the mark by an egg albumin-specific enzyme-linked immunosorbent assay (ELISA). Details of the sample preparation, ELISA procedure, and calculation of ELISA threshold values (i.e., negative controls) are described by Hagler et al. (2014). These ELISA data provided valuable information on the reliability and steadfastness of the protein-marking procedure over the 6-d course of each phase of the study. Specifically, the percentage of protein-marked specimens obtained from the VTC-E (positive controls) and COT-E (negative controls) whole-plant samples over the 6-d duration of the study was determined. Those specimens collected from the VTC-E plots (positive controls) that did not contain egg albumin were deemed to be ‘false negative’. Conversely, those specimens collected from the COT-E plots (negative controls) that contained the mark were ‘false positive’.

Sweep Net Sampling Procedure

The sweep net sampling protocol was designed to provide a relative estimate of the arthropod distribution in the five distinct zones at the study site (i.e., VTC-E, VTC-M, COT-E, COT-M, and BUF). In total, 100 sweep net samples were taken 1, 3, and 6 d after each marking event (black rectangles shown in Fig. 1). The samples on each sample date were pooled (n = 300) for analysis to bolster sample sizes and to simplify data presentation. A single sweep net sample consisted of sweeping the entire 12.2-m length (25 sweeps) of a row in each of the 100 plots between the hours of 0700 and 0900 using a standard 38-cm-diameter canvas net attached to a 0.9-m wooden handle (#7635HS, BioQuip Products Inc., Rancho Dominguez, CA). After each sweep collection, the contents of the sample were placed into a 3.8-liter plastic zip-top bag. The samples were put into a chilled ice chest within a few minutes after each collection and frozen within 1 hr in a −60°C freezer. The sweep samples were processed by removing them from the freezer and carefully searching each sample bag for arthropods. The arthropods were identified to genus or species and tallied to determine the relative estimate of arthropods inhabiting the five sampling zones listed above. These data were also grouped into the three categories described above for data presentation. A listing of the specific species encountered, and their seasonal abundance is provided in Supp Table 1 (online only).

Differences in the abundance of each grouping between the five sampling zones during the early, middle, and late cotton season growth phases were first analyzed by a one-way analysis of variance (ANOVA) using JMP. Again, the arthropod counts did not meet the assumptions needed for a valid ANOVA. Therefore, the count data were analyzed by the nonparametric Kruskal–Wallis rank-sum test to identify significant differences in arthropod populations between the five focal sampling zones. All results are presented as the mean (±SD) number of arthropods collected per sweep sample.

Detection of Protein Mark

The PIT was used to assess the movement of arthropods beyond the protein-marked vernonia trap crop rows (VTC-E) into the various cotton-sampling zones. Each arthropod collected by sweep netting was placed in a 1.5-ml microcentrifuge tube containing 1.0 ml of Tris-buffered saline and examined by ELISA for the presence of the protein mark. The ELISA data yielded from the arthropods collected in the VTC-E zone provided information on the reliability and steadfastness of the marking procedure. Specifically, the percentage of individuals with the egg albumin mark provided a temporal (e.g., 1, 3, and 6 d after marking) estimate of the efficiency of the mark-capture procedure. Distances between the center of the sweep net sampling sites and the nearest possible protein-marked vernonia plots were calculated using the measuring tool in ArcGIS Desktop since the field was planted with high precision techniques. The total number of protein-marked specimens of each arthropod group was tallied. All results for the dispersers are presented as the mean (±SD) distance traveled from the protein-marked vernonia plots.

Results

Whole-Plant Samples

Arthropod Counts

The whole-plant sampling technique provided a per-capita measurement of the number of arthropods inhabiting an individual vernonia (VTC-E) and cotton (COT-E) plant during the early and middle phases of the cotton-growing season. Arthropod density was extremely low during the early growth phase of the cotton season, regardless of the plant type (Fig. 2A). However, there was an indication that predaceous true bugs were beginning to colonize the VTC-E plots. Arthropod densities increased sharply by the middle of the cotton-growing season, and the count data revealed that the various arthropod groups were amassing in the trap crop (Fig. 2B). The true bug pest, true bug predator, and spider densities were approximately 8, 12, and 4 times higher, respectively, in the VTC-E zone than in the COT-E zone. Most importantly, the cotton plants only contained 2.3 (±3.3) pests per plant during the middle phase of the growing season (Fig. 2B).

Fig. 2.

Mean (± SD) number of true bug pests, true bug predators, and spiders per cotton (black bars) and vernonia plant (gray bars). Whole-plant samples were collected during the early (A) and middle (B) growth phase of the cotton season. Symbols above each paired comparison indicate highly significant (P < 0.001***), significant (P < 0.05*), and nonsignificant (ns) differences in arthropod counts between the plant types as determined by the Mann–Whitney two-sample rank-sum test (n = 24 whole-plant samples per plant treatment).

Protein Mark Acquisition and Retention

The arthropods captured by the whole-plant sampling procedure were examined for the presence of egg albumin to assess the fidelity of the mark–capture technique. Overall, only 3 of the 127 (2.4%) specimens collected from the cotton plants yielded a false-positive ELISA reaction (Table 1). Conversely, egg albumin was detected on 924 (95.1%) of the 972 individuals collected from the vernonia plants. There was no discernable decline in marking efficacy over the 6-d duration of the study (Table 1).

Table 1.

ELISA results yielded for the arthropod specimens collected during the early and middle growth phase of the cotton season

Growth phase	Taxon	Days after marking	Sampling zone
			COT-E			VTC-E
			Total collected	No. positive	% positive	Total collected	No. positive	% positive
Early	True bug pest	1	1	0	0.0	3	3	100.0
		3	0	0	na	3	3	100.0
		6	1	0	0.0	8	8	100.0
	True bug predator	1	4	0	0.0	22	22	100.0
		3	4	0	0.0	25	23	92.0
		6	3	0	0.0	28	23	82.1
	Spider	1	2	0	0.0	10	10	100.0
		3	1	0	0.0	5	5	100.0
		6	1	0	0.0	2	2	100.0
Early growth phase totals			17	0	0.0	106	99	93.4
Middle	True bug pest	1	12	1	8.3	148	142	95.9
		3	17	0	0.0	117	107	91.5
		6	25	0	0.0	159	147	92.5
	True bug predator	1	7	0	0.0	100	100	100.0
		3	13	2	15.4	95	91	95.8
		6	8	0	0.0	138	137	99.3
	Spider	1	5	0	0.0	41	37	90.2
		3	14	0	0.0	34	32	94.1
		6	9	0	0.0	34	32	94.1
Middle growth phase totals			110	3	2.7	866	825	95.3
Grand totals			127	3	2.4	972	924	95.1

Whole-plant samples were collected from the cotton (COT-E) and vernonia (VTC-E) sampling zones 1, 3, and 6 d after marking (n = 8 per plant type per day).

Sweep Net Samples

The counts obtained from the sweep net sampling effort provided a relative estimate of arthropod abundance in the five targeted sampling zones. The overall demographics and abundance of the specific arthropod species collected throughout the study are given in Supp Table 1 (online only). It should be noted that 18,710 arthropods were collected over the course of the entire study. The ELISA conducted to detect the presence of the egg albumin mark on the 18,710 arthropods collected by sweep netting provided a means to estimate the migration of the various taxa from the vernonia trap crop (VTC-E) into the cotton field. In addition, the proportion of marked individuals collected in the VTC-E plots provided another estimate of the effectiveness of the protein-marking procedure.

True Bug Pests

In total, 1,530 pest bugs were collected in the 300 sweep net samples taken at the study site during the early phase of the cotton-growing season. The pest bug complex was dominated by the potato leafhopper, Empoasca fabae Harris (Hemiptera: Cicadellidae), and the cotton fleahopper, Pseudatomoscelis seriatus (Reuter) (Hemiptera: Miridae), which accounted for 75% of the pest population (Supp Table 1 [online only]). Pest counts varied from an average of 2.7 (±2.3) in the BUF sampling zone to 13.0 (±5.7) in the VTC-E zone (Fig. 3A). The Kruskal–Wallis rank-sum analysis showed that there was a significantly higher population of pests in the VTC-E (vernonia) sampling zone than in the other zones (df = 4, χ ² = 96.31, P < 0.001; Fig. 3A). In the early phase, 88.4% of the pests collected in the VTC-E plots were marked with egg albumin (Fig. 3A). Of the 1,530 pests collected, 632 (41.3%) were labeled with egg albumin. Five hundred and fifty-one of the 632 marked pests were collected from the trap crop. The net distance traveled by the 81 individual pests that were captured beyond the VTC-E sampling zone was 16.4 ± 27.0 m.

Fig. 3.

Bar charts depicting the mean (±SD) number (left y-axis) of pest bugs collected by sweep netting the various sampling zones (see Fig. 1 for the spatial orientation of the zones) during the early (A), middle (B), and late (C) phase of the cotton-growing season. Overall, 300 sweep samples were taken from the COT-E (n = 48), COT-M (72), BUF (60), VTC-M (72), and VTC-E (48) zones, respectively, for each phase of the study. Different letters above the error bars indicate significant differences in the mean rank of arthropod counts. The red triangles show the percentage (right y-axis) of protein-marked specimens collected in each sampling zone.

The pest bug population tripled by the middle of the cotton-growing season (n = 4,583) with E. fabae and P. seriatus accounting for 87.2% of the population (Supp Table 1 [online only]). The average pest densities from the five sampling zones varied from 9.0 (±3.7) in the COT-E zone to 32.8 (±13.2) in the VTC-E zone (Fig. 3B). Again, the VTC-E pest counts were much higher than in all other zones (df = 4, χ ² = 111.14, P < 0.001; Fig. 3B). Marking efficiency was 63.9% (Fig. 3B). The 213 protein-marked pests collected outside the VTC-E zone traveled an average of 21.2 ± 27.6 m.

The pest bug population continued to rise (n = 6,223) during the late phase of the growing season. This was due to a substantial increase of E. fabae throughout the study site. The average number of pests collected per sweep sample varied from 15.1 (±11.9) in the VTC-M zone to 26.7 (±47.3) in the COT-M zone (Fig. 3C). The high average density of E. fabae in the COT-E and COT-M sampling zones can be attributed to extremely dense aggregations (i.e., outliers) in 9 of the 120 sweep net samples taken in the COT main treatment zone. The analysis of the sweep count data indicated that the pests (majority E. fabae) were amassing in the vernonia (VTC-E) and main COT treatment (COT-E and COT-M) plots (df = 4, χ ² = 12.74, P = 0.013). Marking efficiency during the late phase of the study was 61.3% (Fig. 3C). Only 787 of 6,225 pest bugs collected were marked. Of these, 717 were collected in the VTC-E zone. The 70 protein-marked individuals collected beyond the VTC-E zone traveled an average of 27.6 ± 32.1 m.

True Bug Predators

In total, 2,005 predaceous true bugs were collected during the early phase of the cotton-growing season. The minute pirate bug, Orius tristicolor (White) (Hemiptera: Anthocoridae), and the big-eyed bug, Geocoris punctipes (Say) (Hemiptera: Geocoridae), comprised 80% of the total population (Supp Table 1 [online only]). The minute pirate bug population was about three times higher than the big-eyed bug population. The average predator counts obtained from the five focal sampling zones varied from 4.2 (±3.1) in the COT-E zone to 17.2 (±9.8) in the VTC-E zone (Fig. 4A). The counts were significantly higher (i.e., about four times higher) in the VTC-E plots than in any of the other sampling zones (df = 4, χ ² = 83.68, P < 0.001; Fig. 4A). The marking efficiency for the true bug predators was 88.8% (Fig. 4A). The 56 marked predators collected beyond the VTC-E area traveled an average of 7.1 ± 17.0 m.

Fig. 4.

Bar charts depicting the mean (±SD) number (left y-axis) of predaceous true bugs collected by sweep netting the various sampling zones (see Fig. 1 for the spatial orientation of the zones) during the early (A), middle (B), and late (C) phase of the cotton-growing season. Overall, 300 sweep samples were taken from the COT-E (n = 48), COT-M (72), BUF (60), VTC-M (72), and VTC-E (48) zones, respectively, for each phase of the study. Different letters above the error bars indicate significant differences in the mean rank of arthropod counts. The red triangles show the percentage (right y-axis) of protein-marked specimens collected in each sampling zone.

The total number of predaceous true bugs collected during the middle of the cotton-growing season was 2,547. Again, the predaceous true bug complex continued to be dominated by O. tristicolor and G. punctipes (Supp Table 1 [online only]). However, the predominant species switched to G. punctipes. The average number of predaceous true bugs captured per sample varied from 5.6 (±2.9) in the COT-E zone to 13.0 (±5.5) in the VTC-E zone (Fig. 4B). The predator counts were significantly higher in the VTC-E sample zone than in the other sampling zones (df = 4, χ ² = 74.35, P < 0.001). The marking efficiency during the middle of the cotton season was 81.8% (Fig. 4B). Five hundred and twelve of the 622 protein-marked specimens were collected in the trap crop. The net distance traveled by the 110 marked predators captured beyond the VTC-E zone was 13.6 ± 22.9 m.

There was a substantial decline in the true bug predator abundance during the late phase of the cotton-growing season, with 582 collected. Again, O. tristicolor and G. punctipes were the predominant species and occurring at equal abundance (Supp Table 1 [online only]). Average predator densities in the various sampling zones varied per sweep sample from 0.75 (±0.8) in the COT-E zone to 6.2 (±4.1) in the VTC-E zone (Fig. 4C). Again, the predator population was significantly higher in the VTC-E plots than anywhere else (df = 4, χ ² = 100.59, P < 0.001). Protein-marking efficacy was at 82.4% (Fig. 4C). The marked predators were aggregated in the VTC-E sampling zone. Specifically, 244 of the 269 marked specimens were capture in the vernonia. The 25 marked predators that were captured beyond the VTC-E sampling zone traveled an average of 8.2 ± 18.1 m.

Spiders

In total, 260 spiders were captured during the early growth phase of the cotton season. The crab spider, Mecaphesa celer (Hentz) (Araneae: Thomisidae), and the meshweaver spider, Dictyna reticulata Gertsch and Ivie (Araneae: Dictynidae), were the most abundant taxa comprising 71% of the population (Supp Table 1 [online only]). The average number of spiders collected in the sweep samples varied from 0.1 (±0.3) in the BUF zone to 2.8 (±2.0) in the VTC-E zone (Fig. 5A). There were significantly more spiders in the VTC-E than in any of the cotton-sampling zones (df = 4, χ ² = 102.85, P < 0.001). Two-thirds of the spiders collected from the VTC-E plots were marked (Fig. 5A). Overall, 93 of 260 field-collected spiders were marked with egg albumin. All but four of these spiders were collected in the VTC-E plots. The net distance traveled by the marked spiders captured beyond the VTC-E zone was 7.6 ± 11.2 m.

Fig. 5.

Bar charts depicting the mean (±SD) number (left y-axis) of spiders collected by sweep netting the various sampling zones (see Fig. 1 for the spatial orientation of the zones) during the early (A), middle (B), and late (C) phase of the cotton-growing season. Overall, 300 sweep samples were taken from the COT-E (n = 48), COT-M (72), BUF (60), VTC-M (72), and VTC-E (48) zones, respectively, for each phase of the study. Different letters above the error bars indicate significant differences in the mean rank of arthropod counts. The red triangles show the percentage (right y-axis) of protein-marked specimens collected in each sampling zone.

Spider abundance was about 2.5 times greater during the middle phase of the cotton-growing season than the early phase. Crab spiders comprised 53.7% of the spider population (Supp Table 1 [online only]). The average number of spiders captured in the sweep samples varied from 1.2 (±1.1) in the COT-E zone to 5.6 (±3.2) in the VTC-E zone (Fig. 5B). Spider abundance was significantly greater in the VTC-E than elsewhere (df = 4, χ ² = 93.36, P < 0.001). Marking efficiency was at 77.5% (Fig. 5B). Overall, 227 protein-marked spiders were captured. Of these, 207 were from the VTC-E zone. The average distance traveled by the 20 marked spiders collected beyond the VTC-E was 9.0 ± 20.2 m.

The spider population declined (n = 311) late in the cotton-growing season. Crab spiders continued to be the predominant spider taxon, comprising 43% of the population (Supp Table 1 [online only]). The average capture rate of spiders varied from 0.6 (±0.9) in the COT-E zone to 2.9 (±2.8) in the VTC-E zone (Fig. 5C). Again, there were significantly more spiders in the VTC-E plots than in any of the cotton-sampling zones (df = 4, χ ² = 36.58, P < 0.001). The late-season marking efficiency for the spiders was 82.0% (Fig. 5C). Overall, 132 of the field-collected spiders were marked with egg albumin. Of these, only nine were captured beyond the VTC-E, all of which were captured in the nearest VTC-C plot. The distance traveled by these specimens was only 2.5 ± 0.9 m.

Discussion

The distribution of arthropods inhabiting the vernonia trap-cropped cotton field was assessed by whole-plant and sweep net sampling procedures. The whole-plant sampling technique provided an absolute count of arthropods on a per plant basis (Ellington et al. 1984, Hagler and Blackmer 2013). The per-plant arthropod census data clearly showed that they preferred vernonia over cotton. It should be noted that whole-plant samples were only taken from the VTC-E and COT-E plots. This was because the whole-plant sampling technique is very labor intensive and time consuming.

The sweep net sampling procedure provided a faster and less labor-intensive method to obtain relative estimates of arthropod abundance throughout the entire study site (Spurgeon 2009, Hagler et al. 2018). To provide perspective, we collected 18,710 arthropod specimens in the 900 sweep samples (25 sweeps per sample) taken over the course of the cotton season. The sweep net census data confirmed that the arthropods were strongly aggregated in the 4.2% of the field space dedicated to vernonia. The sweep net sampling technique was also useful for determining the dispersal rate of arthropods from the trap crop into the cotton. Every field-collected specimen was examined by ELISA for the presence of the egg albumin mark. The mark was detected on 25.4% (n = 4,757) of the field-collected arthropods. Of these, 88% (n = 4,169) were captured in the VTC-E plots (i.e., nondispersers). These data indicate that vernonia retained the arthropods throughout the study. Protein-marking efficiency was very high, with egg albumin detected by ELISA on 95 and 74% of the arthropods collected in the VTC-E plots using the whole-plant and sweep net sampling schemes, respectively. The discrepancy in marking efficacy between the two sampling techniques was expected. The arthropods trapped in the whole-plant enclosures were more exposed to residual contact with the protein-marked vernonia plant or to potential protein contamination of the netting surrounding each marked plant. Also, the cages eliminated the possibility of unmarked arthropods entering the sampling unit. Previous protein mark–capture studies using whole-plant sampling procedures have reported similarly high marking efficiencies (Hagler et al. 2014, Hagler and Machtley 2020). Sweep net sampling, meanwhile, provided a more realistic temporal assessment of marking efficiency under open field conditions. Given the high marking efficiency of caged specimens, we can be confident in assuming that the lower percentages of marked specimens obtained from the sweep net samples taken in the VTC-E plots was most likely due to the collection of unmarked arthropods that had emigrated from the cotton or another nearby crop.

The arthropod counts were pooled into three broad categories (i.e., true bug pests, true bug predators, and spiders) for data presentation. Our rationale for combining the species into these broad groups was that the ‘key’ species in our study suffered from relatively low counts (i.e., zero counts). Specifically, the original focus of the study was to examine the population dynamics of the western tarnished plant bug, L. hesperus, in this trap-cropped system. Lygus hesperus is a perennial pest at this research site (Hagler and Naranjo 1994; Naranjo et al. 2003, 2004; Hagler and Blackmer 2013). However, only 560 adults were captured by sweep netting over the entire course of this study (Supp Table 1 [online only]). Sweep net counts did show that L. hesperus were, in general, four times higher in the VTC-E plots than in the cotton plots.

The most abundant pest encountered during this study was the potato leafhopper, E. fabae. The potato leafhopper population steadily increased over the course of the study. In total, 116, 676, and 5,043 E. fabae were collected during the early, middle, and late phases of the cotton-growing season, respectively (Supp Table 1 [online only]). Cotton is a known host plant for this polyphagous pest. However, it is also known that potato leafhoppers prefer other crops, such as alfalfa (Lamp et al. 1994, Chasen et al. 2014). The prevalence of the potato leafhopper was unexpected as it has been absent from our previous arthropod surveys at the MAC (Hagler and Naranjo 1994; Naranjo et al. 2003, 2004; Hagler and Blackmer 2013). It is possible that vernonia served as the source host plant for the potato leafhopper outbreak in the cotton, but evidence suggests otherwise. First, the overall E. fabae density was lower in the main VTC treatment zone (n = 2,169) than in the comparable main COT treatment zone (n = 2,839; Supp Table 1 [online only]). Second, the low frequency of marked leafhoppers captured beyond the VTC-E sample plots supports the hypothesis that it was not moving from the vernonia. The probable source for the potato leafhopper invasion was from two nearby alfalfa fields located <200 m from east and southwest our field site. The investigator conducting research on the alfalfa revealed that there was a high density of E. fabae in his fields (A. Mostafa, personal communication). The late-season surge in potato leafhopper density was likely due to a mass dispersal from the alfalfa after it was harvested. Although the E. fabae abundance was high, there was no noticeable damage caused to the cotton.

The other abundant pest encountered was the cotton fleahopper, P. seriatus. Like L. hesperus, this mirid is a cotton pest because it tends to feed on the young cotton bolls (squares). Previous studies at MAC have shown that L. hesperus populations are typically 4–10 times more abundant than P. seriatus (Naranjo et al. 2003, 2004; Naranjo 2005). In the present study, P. seriatus density (n = 4,038) was over seven times greater. It is possible that the relatively high fleahopper density was due to the presence of the vernonia trap crop. However, it should be noted that over half (51%) of the fleahoppers were collected in the VTC-E sampling plots or in cotton-sampling plots nearest (e.g., rows 3 and 10 in spans 7 and 8; see Fig. 1) to the vernonia. As such, the cotton fleahoppers were aggregated in the vicinity of the trap crop and never reached pest status in the cotton.

The most abundant predators encountered over the course of the study were the big-eyed bug (G. punctipes), minute pirate bug (O. tristicolor), and crab spider (M. celer). Historically, these species have been among the most predominant predators encountered at the MAC farm (Naranjo et al. 2003, 2004; Hagler and Blackmer 2013; Hagler et al. 2018). The whole-plant and sweep net census data clearly showed that the predators were amassing in, or the near vicinity of, the -E sample plots. The aggregations of these generalist omnivores in vernonia could be due to a combination of two factors. First, they could be congregating in response to both a greater diversity and density of available prey in vernonia (Stern et al. 1969, Hogg et al. 2011). Second, they may have a stronger host plant affinity for vernonia. Previous research has shown that the omnivorous pest L. hesperus is attracted to blooming vernonia plants (Hagler et al. 2016). Other studies have shown that G. punctipes, O. tristicolor, and M. celer also engage in phytophagy or nectivory (Naranjo and Gibson 1996, Taylor and Pfannenstiel 2008, Lorenzo et al. 2019). In short, it is possible that these predators were also attracted to vernonia’s floral rewards (Balmer et al. 2013).

Holden et al. (2012) developed a simple simulation model to identify the critical components of a useful trap crop, which indicated that the long-term retention of the pests (and natural enemies) in the trap crop might be a crucial factor. Count data obtained from our sampling effort revealed that vernonia supported a higher population of arthropods than did cotton, and the PIT mark–capture data showed that the arthropods were well retained in the trap crop throughout the cotton-growing season. A high proportion of arthropods collected from the trap crop (73.8%) tested positive, strongly suggesting that there was not a high rate of turnover. There was a measurable movement of both pests and natural enemies from the trap crop into the adjacent cotton, albeit at a low level (4.5% of arthropods recovered from cotton were marked). The omnipresent floral bloom emitted by vernonia was probably the critical factor for the strong attraction and retention by the arthropod complex. However, it is impossible to determine exactly what proportion of arthropods from the trap crop dispersed into the adjacent cotton, as opposed to other agricultural fields in the surrounding area. A future study tailored to precisely quantify turnover should be conducted using the PIT to determine this type of interaction.

As Holden et al. (2012) stated, the major benefit of having an enduring trap crop is that it provides avenues for other types of pest management that are economically and environmentally sound. The most apparent benefit, without causing any extra field maintenance, would be the creation of a refuge for natural enemies. Ultimately, this should enhance the biological control services rendered by the natural enemy community (Godfrey and Leigh 1994, Hagler et al. 2018). Although some studies have failed to find such a benefit from flower plantings (Tscharntke et al. 2016, Quinn et al. 2017), there have been many reported successes (Tschumi et al. 2015, 2016; Campbell et al. 2017, Albrecht et al. 2020). In our present study, L. hesperus density never reached its action threshold for control at our study site (Ellsworth 2000). As such, no pesticides were applied for Lygus spp. or any other cotton pest. It should be noted however, that in other nearby research and commercial cotton fields which did not contain a trap crop, L. hesperus did reach its action control threshold and had to be treated with pesticide.

Possible complimentary pest control tactics, which would require various levels of field maintenance, include 1) the mechanical removal of pests aggregated in the trap crop by vacuuming (Swezey et al. 2007), 2) trapping of pests with colored sticky cards (Moreau and Isman 2011), 3) retaining pests and natural enemies in the trap crop with semiochemical lures (Tillman and Cottrell 2012, Kelly et al. 2014), and 4) application of pesticides only to the pest aggregations in the trap crop (Cavanagh et al. 2009).

A vernonia possesses many characteristics of an effective trap crop, more research is needed before it is deployed. The optimal spatial and temporal placement of the trap crop in the field has yet to be identified. Indeed, our rationale for sowing the seeds on two separate dates (January and February) was that we were unsure of the germination and growth rates. The germination rate of both cohorts was excellent, and the young seedlings flourished in the greenhouse environment. We then transplanted both cohorts in the field in mid-April. The older plants started to bloom by the middle of June. Subsequently, there was a continuous floral bouquet of vernonia until early August. The prolonged and synchronous blooming period of the two vernonia cohorts was ideal because, as stated above, the arthropods had an affinity for the flowering plants. The net result is that pest densities were very low in the cotton during its most vulnerable stage to pest attack.

However, there was a caveat to the relatively early blooming vernonia plants. Specifically, those plants began to mature and shed their seeds in late July. In turn, the seeds germinated in August. This raised concern that the second-generation ‘volunteer’ plants could become a weedy pest in cotton or other nearby fields. Adding to the concern about weed potential, vernonia is prolific at producing seed (JRH, personal observation). Moreover, its seed is highly adapted for dispersal, being light and having a feathery pappus. The pappus is a modified calyx that is common to plants in the Asteraceae family. It serves as a ‘parachute’ that facilitates seed dispersal through wind and water (e.g., rainfall, irrigation canals, etc.) and attachment to animals.

Clearly, research is needed to mitigate the unintended propagation of vernonia. Perhaps the seed could be planted directly in the field in March instead of transplanting 1- and 2-mo-old seedlings. Then, the vernonia could be terminated (e.g., by plowing) later in the season after the cotton is no longer vulnerable to true bug pests and before vernonia seed maturation. Another option would be to spatially and temporally stagger vernonia plantings throughout the cotton field in a manner that resembles the alfalfa strip-cutting trap crop method (Stern et al. 1964, Godfrey and Leigh 1994). Then, as the early-season plants mature, they could be plowed. This tactic would mitigate seed propagation and ensure a constant floral bloom is present in the field.

From a spatial perspective, research to date has shown that about 10% of the crop space is usually dedicated for a typical trap crop system (Hokkanen 1991, Shelton and Badenes-Perez 2006). Here, we arbitrarily chose to devote four rows of vernonia to the 96-row cotton field (i.e., 4.2% of the field). Moreover, these four rows were skewed toward one edge of the field. This spatial placement was chosen to create a trap crop and nontrap crop (control) treatment zone in the limited amount of space that was available at the study site. We acknowledge that this was probably not an optimal spatial orientation. It is highly likely that other spatial arrays of the trap crop would yield even better results. For instance, equidistant plantings of individual rows of vernonia throughout the cotton field would be a logical planting scheme (i.e., a 1:23 vernonia to cotton row planting ratio). Another reasonable trap crop array would be to place the vernonia along the borders of the field (Root 1973, Hunt and Whitfield 1996, Aluja et al. 1997, Boucher et al. 2003, Potting et al. 2005, Tillman et al. 2009, Blaauw et al. 2017) or in unused spaces near the cotton crop (e.g., along roadsides, in fallow areas, etc.).

A future option for utilizing the ecosystem services provided by vernonia as a trap crop could be to grow it adjacent to cotton as a commercial crop. Given that vernonia is a heat-tolerant crop requiring little agronomic input, it could be an ideal companion crop for cotton growers. Vernonia yields high quantities of epoxy fatty acids useful in the reformulation of oil-based paints to reduce the emission of volatile organic compounds that contribute to smog production (Perdue et al. 1986). Other potential markets for the fatty acids include eco-friendly plasticizers, additives in PVC, polymer blends and coatings, and cosmetic and pharmaceutical applications (Carlson and Chang 1985). However, as agronomic research on vernonia is relatively early (Shimelis and Gwata 2013), it may be years before this option becomes commercially viable.

Along with attracting pests and predators, it is worth noting that the blooming vernonia provided a summer habitat for a wide variety of pollinators that are not often seen in a cotton monoculture (JRH, personal observation). There was a continuous flow of honey bees, bumble bees, native bees, butterflies, wasps, and flies in the vernonia throughout the growing season (JRH, personal observation). Unfortunately, very few of these pollinators were captured in either of our sampling protocols; thus, their abundance and aggregation patterns are not reflected in our data. The potential conservation of pollinators and the ecosystem services they provide in the presence of vernonia deserve further investigation.

In summary, this study showed that vernonia possesses many of the characteristics of an ideal trap crop or flower strip. True bug pests and predaceous natural enemies that are endemic to cotton aggregated and attenuated their movement in the 4.2% of the farm space dedicated to the vernonia trap crop. Given the aggregation of pests in the trap crop, and that the cotton field did not require pesticide treatment during our study period, there appears to be real potential for vernonia to provide ecosystem services to cotton fields. Also, vernonia did well planted in rows alongside cotton and irrigated, fertilized, and cultivated for weeds on the same schedule as cotton. More research is needed however, on the spatiotemporal placement of the trap crop, particularly concerning whether vernonia could become a weedy pest.

Author Contributions

J.R.H.: conceptualization; experimental design, data curation; data analysis; resources; supervision; visualization; writing original draft. A.L.T.: methodology; investigation, editing original draft. S.A.M.: data curation; investigation; methodology; data acquisition; editing original draft. M.T.C.: data curation; investigation; methodology; data acquisition; editing original draft.