A Molecular Approach for Detecting Stage-Specific Predation on Lygus hesperus (Hemiptera: Miridae)

Abstract

A molecular gut analysis technique is described to identify predators of Lygus hesperus (Knight), a significant pest of many crops. The technique is unique because it can pinpoint which life stage of the pest was consumed. Sentinel egg masses designed to mimic the endophytic egg-laying behavior of L. hesperus were marked with rabbit serum, while third instar and adult L. hesperus were marked with chicken and rat sera, respectively. Then, the variously labeled L. hesperus life stages were introduced into field cages that enclosed the native arthropod population inhabiting an individual cotton plant. After a 6-h exposure period, the predator assemblage, including the introduced and native L. hesperus population, in each cage were counted and had their gut contents examined for the presence of the variously marked L. hesperus life stages by a suite of serum-specific enzyme-linked immunosorbent assays (ELISA). The whole-plant sampling scheme revealed that Geocoris punticpes (Say) and Geocoris pallens Stal (Hemiptera: Geocoridae) and members of the spider complex were the numerically dominant predator taxa in the cotton field. The gut content analyses also showed that these two taxa appeared to be the most prolific predators of the L. hesperus nymph stage. Other key findings include that Collops vittatus (Say) (Coleoptera: Melyridae) and Solenopsis xyloni McCook (Hymenoptera: Formicidae) appear to be adept at finding and feeding on the cryptic L. hesperus egg stage, and that L. hesperus, albeit at low frequencies, engaged in cannibalism. The methods described here could be adapted for studying life stage-specific feeding preferences for a wide variety of arthropod taxa.

The western tarnished plant bug, Lygus hesperus Knight, is regarded as one of the most persistent and destructive cotton pests in the United States (Stern et al. 1964, Scott 1977, Young 1986, Hagler et al. 1991). Considerable effort has been taken to identify predators of L. hesperus in cotton and other agroecosystems (Leigh and Gonzalez 1976; Hagler et al. 1992, 2018a; Zink and Rosenheim 2008; Hagler 2011; Hagler and Blackmer 2013). A better knowledge of which predators are exerting the greatest biological control pressure on L. hesperus at the species level, and the life stage level, will enhance IPM programs dedicated to this pest.

Many generalist predators that inhabit cotton have been identified, via postmortem gut content analysis, as natural enemies of L. hesperus. Thirty years ago, a monoclonal antibody (MAb)-based enzyme-linked immunosorbent assay (ELISA) was developed that detected L. hesperus egg protein remnants in the guts of predators (Hagler et al. 1991, 1992). This assay could detect egg and gravid female predation events, but it was ineffective at detecting remnants of immatures and adult males. Subsequently, a L. hesperus-specific polymerase chain reaction (PCR) assay was developed as a tool for gut analysis (Hagler and Blackmer 2013, Hagler et al. 2018a). The PCR assay was reliable at detecting L. hesperus DNA of all the life stages in predators, but it was ineffective at deciphering which life stage was consumed.

A third gut analysis method, recently coined as the universal food immunomarking technique (UFIT), is another valuable tool for biocontrol researchers because of its versatility, ease of use, and accuracy (Hagler 2019). The UFIT consists of tagging the prey item of interest with a unique protein mark. In turn, any predator that consumes a marked prey item can be identified by a standardized (universal) protein-specific ELISA (Hagler and Durand 1994). Two UFIT ELISAs (using anti-rabbit and anti-chicken IgGs) have been used to pinpoint life stage-specific feeding events on Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) eggs and larvae (Mansfield et al. 2008, Rendon et al. 2018) and L. hesperus eggs and nymphs (Hagler 2006, 2011; Hagler and Mostafa 2019). These UFIT ELISAs have also been used to distinguish between a predator’s preference for live prey or carrion (Mansfield and Hagler 2016), an omnivore’s preference for plant material or live prey (Blubaugh et al. 2016), and a predator’s proclivity to engage in intraguild and intraspecific (i.e., cannibalistic) predation (Hagler et al. 2020).

In this paper, we describe how the UFIT can be easily modified to simultaneously identify predators of all L. hesperus life stages (e.g., egg, nymph, and adult). To achieve this goal, we: 1) marked L. hesperus eggs, nymphs, and adults with specific protein labels, 2) released the marked prey into field cages that contained the native cotton predator complex, and 3) subsequently assayed the predators for the presence of protein-marked prey remnants using a suite of serum-specific ELISAs. We also introduce a third ELISA (anti-rat serum) to the suite of assays available for UFIT research. The combined whole-plant sampling (cage) and UFIT methods described here yielded data on per-capita estimates of the predator population in the cotton field and precisely identified which predators were feeding on the various L. hesperus life stages.

Materials and Methods

Prey Marking Procedures

The L. hesperus used in the study were obtained from a colony reared on the artificial diet described by Debolt (1982). Third-instar and adult L. hesperus were marked internally by feeding them this diet to which was added a solution of 10-mg/ml chicken serum (Millipore Sigma #C5405, St. Louis, MO) and rat serum (bioWORLD #30611164, Dublin, OH), respectively. The protein-laced diets were supplied to the specimens starting at their second instar. The day before their release into the field cages, all individuals were externally marked; nymphs were sprayed with 1.5 ml of 10-mg/ml chicken serum solution with a hand-pump spray bottle, whereas adults were sprayed with 2 ml of 10-mg/ml rat serum with an airbrush, both in vented plastic containers. The L. hesperus eggs were marked as described in detail by Hagler and Mostafa (2019). To briefly summarize, egg packets consisting of agarose gel enclosed in Parafilm (Bemis Company, Inc., Neenah, WI) were removed from the laboratory colony. The eggs embedded in the Parafilm were cut into narrow strips and gently peeled away from the underlying agarose gel to expose the eggs entirely. Each piece contained ~40 eggs. Each exposed egg was marked by stroking with a fine-tipped paintbrush dipped into a 10-mg/ml solution of rabbit serum (Fisher Scientific #16120099, Pittsburgh, PA). The Parafilm strip containing the rabbit serum-marked eggs was wrapped around a wood dowel (serving as an artificial cotton stem). The dowel had a notch cut around its circumference, so the eggs could be placed facing inward, simulating the characteristic endophytic oviposition behavior of L. hesperus by concealing the eggs in the notch of the artificial stem. A diagrammatic depiction of the artificial oviposition substrate is given in Hagler and Mostafa (2019).

Field Site

The experiment was conducted within a 0.67-ha cotton field located at the University of Arizona Maricopa Agricultural Center, Maricopa, AZ. This field was planted between several other 0.67-ha cotton fields.

Field Cage Arenas (Whole-Plant Sampling Procedure)

Whole-plant field cages were placed on randomly selected cotton plants on 9 August and 30 August 2019. In total, 22 cages were erected on each day. The caging technique, as described by Hagler (2011), is designed to capture the natural arthropod community inhabiting a single cotton plant while, at the same time, containing the protein-marked L. hesperus nymphs and adults introduced into each cage. Each cage consisted of a 1.0-m long × 0.5-m diameter organdy (fine mesh fabric) tube-shaped sleeve. Two days before the initiation of the predation study, the bottom opening of each cage was sealed around the base of a single cotton plant with a nylon zip-tie. The top of each cage was left open and unrolled at ground level to allow the natural arthropod fauna to re-inhabit the disrupted plants.

At the start of the study, each cage was rapidly pulled up over each plant to capture the native insects, including any predators or L. hesperus nymphs and adults, on each plant. Keeping the top opening of the cage as narrow as possible, a wood dowel containing a strip of rabbit serum-marked eggs, as described above, was attached to the main cotton stem with a binder clip. Then, three third-instar L. hesperus marked with chicken serum and three adults marked with rat serum were released into each cage. All releases were made between 06:00 and 07:00. The top of each cage was closed with a wire twist tie to block the movement of arthropods in and out of the cages. All arthropods enclosed in each cage were allowed to move and forage freely within for 6 h. Note that previous research showed that protein-marked L. hesperus eggs, nymphs, and adults could be detected reliably by ELISA in predators for at least 4, 6, and 6 h, respectively, after a meal (Hagler 2006, Hagler and Mostafa 2019). After 6 h, each caged cotton plant was cut at its base just below the zip-tie, placed in a plastic garbage bag, and frozen within 30 min of collection in an ultra-low freezer (−80°C) to quickly kill the arthropods in each cage.

Each whole-plant sample was processed by removing from the freezer and carefully searching for predators and L. hesperus nymphs and adults. The predators captured in each cage were identified and counted. Each individual was placed into a 1.5-ml microcentrifuge tube and homogenized in 1.0 ml of TBS buffer with a tissue grinder. Between 8 and 10 L. hesperus eggs were collected from the artificial egg patch in each cage. The eggs were removed from the Parafilm, placed individually into a microcentrifuge tube containing 1.0 ml of buffer, and homogenized. Separate disposable toothpicks were used to sort and move all samples into tubes to minimize any protein contamination between samples. The individual L. hesperus egg specimens were examined to check the fidelity of the marking procedure.

Predator Gut Analysis

Each homogenized predator sample was examined for the presence of marker protein carried by each prey. The captured L. hesperus (nymphs and adults) samples were also examined to distinguish between the released (i.e., those yielding a positive reaction) and the native (i.e., those yielding a negative reaction) population. These examinations also served to determine the proclivity for the nymph and adult stages engaging in cannibalism (i.e., a specimen that scored positive for the presence of both marks was assumed to have engaged in a cannibalistic feeding event).

A 100-µl aliquot of arthropod homogenate was used for each ELISA. The well-established anti-rabbit and anti-chicken sandwich ELISAs are described in detail by Hagler and Durand (1994) and Hagler (1997). For the anti-rat sandwich ELISA, the wells of a 96-well microplate (Corning Falcon #351172, Corning, NY) were coated with 100 µl of goat anti-rat antibody (Millipore Sigma #SAB3700561) diluted 1:500 in TBS and incubated for 1 h at 27°C. The antibody was discarded, and 360 µl of soymilk was added for 30 min at 27°C to block the remaining antigen binding sites in each well. The soymilk was discarded and a 100-µl aliquot of the homogenized insect sample was placed in an individual well of the microplate. The samples were incubated for 1 h at 27°C, discarded, and the wells were washed two times with PBS Tween 20 (0.05%) using an automated microplate washer (BioTek Instruments, Winooski, VT). A 50-µl aliquot of anti-rat antibody conjugated to horseradish peroxidase (# SAB3700637, Sigma Chemical Co., St. Louis, MO) diluted to 1:1,000 in soymilk, was added to each well for 1 h at 27°C. The wells were washed as described above and 50 µl of TMB Microwell One Component Peroxidase Substrate (Surmodics #TMBW, Eden Prairie, MN) was added to each. After a 10-min incubation, the optical density (OD) of each well was measured with a SpectraMax iD3 microplate reader (Molecular Devices, San Jose, CA) set at a 405-nm wavelength. Negative (unmarked) arthropod samples, TBS buffer controls, and positive serum controls were included on each ELISA plate.

Negative Control Specimens

Unmarked (negative control) specimens of each taxa were examined to determine the inherent background noise associated with each anti-serum ELISA. The ELISA OD readings yielded by the unmarked specimens of each taxa were consistently low. Therefore, all negative control data were pooled for the critical threshold value (CTV) calculation for each type of ELISA. The calculated CTV value yielded for the anti-rabbit, anti-chicken, and anti-rat gut content ELISA was 0.076, 0.075, and 0.083, respectively. A specimen was scored positive for the presence of protein-marked remnants if its ELISA OD value was greater than twice the average value of the CTV.

Data Analysis

The predator counts of the 10 most predominant taxa collected on the two sample dates were pooled (n = 44 cages) to simplify data presentation. The raw data showing the spatial (between cages) and temporal (between sample dates) differences in predator abundance and feeding activity are provided in Supp Table 1 (online only). Descriptive statistics were calculated to determine the average (±SD) number of each predator taxon collected per plant. Then, the number of predators containing marked L. hesperus egg, nymph, and adult remnants in their guts was determined by tallying those that yielded a positive rabbit, chicken, and rat gut content ELISA reaction. The L. hesperus nymphs and adults were also assayed to detect for cannibalistic feeding activity.

Results

In total, 396 L. hesperus eggs (≈8 to 10 from each cage) were randomly sampled from the egg patches at the end of the 6-h prey exposure period and examined for the presence of the rabbit serum mark. Of these, 368 (94%) contained the targeted mark (Table 1). Of the free-roaming L. hesperus, 59 of 78 (76%) nymphs and 67 of 118 (57%) adults tested positive for the presence of their respective marks (Table 1). We considered unmarked nymphs and adults indigenous to the study site. As such, the average number of native nymphs and adults per cotton plant was 0.4 ± 0.6 and 1.2 ± 1.1, respectively.

Table 1.

The total number of Lygus hesperus eggs, nymphs, and adults collected from the 44 field cages and the total number of specimens testing positive by ELISA for the presence of the egg (rabbit), nymph (chicken), and adult (rat) markers

		Anti-serum ELISA
Lygus hesperus life stage	Number collected	Egg	Nymph	Adult
Egg	396	368	0	0
Nymph	78	1	59	1
Adult	118	2	3	67

Lygus hesperus nymphs and adults were also examined for the presence of each mark to determine their proclivity for cannibalism. The gut content examinations revealed one egg and one adult cannibalism feeding event by the nymphal stage, and two egg and three nymph cannibalism events by the adult stage (Table 1).

The focal predator assemblage consisted of the 10 most abundant taxa encountered during the study. The spatial and temporal differences in predator abundance and feeding activity are given in Supp Table 1 (online only). Overall, an average of 34.5 (± 17.2) predators were collected per cotton plant (Table 2). The gut analyses revealed that 8.9% (68 out of 760) of the predators contained one of the protein marks carried by the introduced L. hesperus in their guts. Of those, 44, 20, and 4 showed evidence of having consumed a protein-marked egg, nymph, and adult, respectively (Table 2).

Table 2.

The mean (±SD) number of predators collected per plant (n = 44 whole plant samples), sum total of individuals collected, number of individuals containing protein-marked Lygus hesperus egg, nymph and adult remnants, and sum total and percent of the population containing remnants

						Lygus hesperus life stage consumed
Class	Order	Family	Predominant taxon	Mean (±SD) per plant	No. collected	Egg	Nymph	Adult	Total consumed (%)
Insecta	Hemiptera	Geocoridae	Geocoris spp.	6.50 (8.00)	286	15	6	1	22 (7.7)
		Anthocoridae	Orius spp.	0.20 (0.55)	9	0	0	0	0 (0)
		Nabidae	Nabis alternatus	0.20 (0.41)	9	0	0	0	0 (0)
		Reduviidae	Zelus renardii	1.50 (1.34)	66	0	1	0	1 (1.5)
	Coleoptera	Melyridae	Collops vittatus	0.48 (0.90)	21	10	1	0	11 (52.4)
		Coccinellidae	Hippodamia convergens	1.40 (1.44)	63	2	2	1	5 (7.0)
			Coccinella septempunctata	0.48 (0.82)	21	1	0	0	1 (4.8)
			Scymnus spp.	1.8 (2.44)	79	0	1	0	1 (1.3)
	Hymenoptera	Formicidae	Solenopsis xyloni	1.36 (3.64)	60	15	0	0	15 (25.0)
Arachinida	Araneae	Thomisidae	Misumenops celer	3.32 (2.05)	146	1	9	2	12 (8.2)
Totals				34.54 (17.18)	760	44	20	4	68 (8.9)

The big-eyed bugs, Geocoris punctipes and Geocoris pallens, comprised 38% (n = 286) of the total predator population, yielding an average of 6.5 (±8.0) individuals per plant. The gut analyses detected protein-marked L. hesperus remains in 7.7% (n = 22) of their population. Of these, 15, 6, and 1 showed evidence of preying specifically on the egg, nymph, and adult life stage, respectively. The next most abundant predator encountered was various species of spiders (20% of the population). The spider complex was dominated by the crab spider, Misumenops celer (Hentz) (Araneae: Thomisidae). Twelve of the 146 (8.2%) spiders examined contained a protein mark in their guts. Of these, 1, 9, and 2 showed evidence of having fed on an egg, nymph, and adult, respectively. The UFIT gut analyses also detected protein in 11 of the 21 Collops vittatus (Say) examined. Of these, 10 contained rabbit serum, the marker for eggs (Table 2).

Another discovery from this study was that fire ants, Solenopsis xyloni McCook, were only encountered on the second sampling date. Moreover, only a few of the field cages were invaded on 30 August (see Supp Table 1 [online only]). Overall, 15 of the 60 (25%) fire ants examined tested positive for the rabbit protein only, suggesting exclusive feeding on the L. hesperus egg stage. It should also be noted that, in some instances, we directly observed the ants foraging on the L. hesperus egg patches. However, our data for ants are limited because they were observed as being able to freely enter and leave several plant cages, presumably through small unintentional gaps.

Discussion

The UFIT has many characteristics that make it a practical tool for predator gut content analysis. First, the protein detection assays have already been developed and proven effective (Hagler and Machtley 2016, Hagler 2019). Therefore, investigators do not need to spend time and resources developing and optimizing the gut assays. Second, the sandwich ELISA format is simple, inexpensive, and well-suited for mass throughput (Sheppard and Harwood 2005, Fournier et al. 2008). Hundreds of predator ELISA samples can be processed in a day at a fraction of the cost of the prey-specific PCR assay. Also, a previous study showed that the sandwich ELISA format used for UFIT research is more reliable than a prey-specific PCR assay (i.e., the UFIT yields fewer false-negative reactions; Hagler et al. 2015). Third, the UFIT can be used to pinpoint predation events at the species and life stage-specific level (Hagler 2019). Here, we showed how it could be modified to identify key predators of each L. hesperus life stage. Such life stage-specificity has been unattainable using the PCR gut analysis approach (Sheppard and Harwood 2005). We also showed that the UFIT could be used to identify the proclivity for L. hesperus to engage in cannibalism.

The whole-plant field cage methodology is also an excellent sampling tool for use in UFIT research. The caging of individual plants serves to trap and contain the arthropods inhabiting a single cotton plant, to hold the protein-marked specimens introduced into the microcosm, and to provide an absolute sampling method for the experiment (Hagler 2006, 2011). The whole-plant sampling procedure revealed spatial and temporal fluctuations in the population counts of the various predator taxa (see Supp Table 1 [online only]). For example, Geocoris spp. was the predominant taxon encountered over the entire course of the study. However, it was 5.5 times more abundant on 9 August than it was on 30 August. Also, some predator taxa were absent on one of the sampling dates (i.e., no Orius tristicolor (White) (Hemiptera: Anthocoridae) or Nabis alternatus Parshley (Hemiptera: Nabidae) were captured on 30 August). The most notable spatial and temporal population fluctuation occurred with the native fire ant, S. xyloni. Specifically, it was absent from the whole plant sampling units on the first sample date (9 August). Moreover, on 30 August, the majority of the 60 ants collected were obtained from 5 of the 22 cages (Supp Table 1 [online only]). It was apparent (by direct focal observation) that foraging ants were entering these cages after they had been closed, finding the L. hesperus egg patches, and subsequently recruiting their nestmates to the food source. This study underscores that S. xyloni has an uncanny ability to invade the field cage enclosures. They will exploit any tiny opening in the cage or even chew holes in the fabric (pers. obs.). Previously, Hagler (2006) modified his experimental design to include an ant cage inclusion/exclusion component to the study. Specifically, ant marauders were excluded from half of the field cages by smearing an insect barrier (i.e., Tanglefoot) around the base of caged cotton plants. This allowed for a direct comparison of the impact that the presence or absence of ants had on the targeted protein-marked prey taxa. This tactic should be considered for future studies (e.g., an ant inclusion and an ant exclusion treatment). A potential caveat of the gut content analysis results yielded by the ants is that they often share their prey with nestmates via trophallaxis. In such a case, a single protein-marked L. hesperus egg could be passed along to nestmates. The net result would be an overestimation of the ant’s impact as an egg predator. This type of food chain error is a well-known, but unavoidable limitation with any molecular gut content analysis procedure (Harwood et al. 2001, Hagler 2016).

The capability to detect and quantify life stage-specific predation is important because pests inflict varying levels of damage at various stages. Zink and Rosenheim (2005, 2008) showed that preferential feeding on different Lygus spp. stages by biocontrol agents can affect the population structure of the pest as well as crop damage. Our method may prove to be a particularly valuable tool in measuring such interactions, as the nymphal stages of Lygus spp. are prone to undercounts in traditional sweep net sampling (Zink and Rosenheim 2004, Musser et al. 2007), and therefore, it may be more difficult to directly measure or estimate nymphal population response to predation pressure.

A key insight obtained from the gut analysis data was the determination of the egg stage being the most vulnerable to predator attack. Overall, the anti-rabbit ELISA detected the presence of the egg marker in 44 out of the 760 individuals examined. The predator taxon-specific results showed that C. vittatus (with one exception) and S. xyloni appeared to prey exclusively on the egg stage. Moreover, the data add support to a view of these taxa as prolific egg predators, as the anti-rabbit ELISA detected the protein in 52 and 25% of the C. vittutus and S. xyloni examined, respectively. As mentioned previously, it is well known that ants often recruit nestmates to a food source (Tschinkel 2011). This study and others (Zilnik and Hagler 2013, Hagler and Mostafa 2019) suggest that C. vittatus are not only adept at finding L. hesperus eggs but they also appear to signal their findings to conspecifics (JRH, pers. obs.).

The anti-chicken ELISA detected 20 presumed predation events on L. hesperus nymphs. Of these, 75% (15) were attributed to spiders and Geocoris spp. Previous research has shown that Geocoris spp. is an important predator of Lygus spp. nymphs (Leigh and Gonzalez 1976, Zink and Rosenheim 2008). Spiders, especially immatures, show more success preying on wingless Lygus spp. nymphs than on the larger, flight-capable adults (Young 1989). It may be that this pattern holds true across all predatory arthropods—the anti-rat ELISA, detecting marked adult L. hesperus, detected the fewest total potential predation events (4), and in no case did our data show a predator feeding on adults more than on nymphs. Even in the context of such a pattern, Geocoris and the spiders stand out as particularly voracious feeders on nymphs. Additional research could reveal more details, such as whether other predators take on a similar role at other points in the season.

One last point of interest was the apparent inter-stage cannibalism shown by L. hesperus. While this species is a documented omnivore (Wheeler 1976, Hagler et al. 2004) and is known to engage in cannibalism in high-density artificial rearing cages (pers. obs.), the frequency of this behavior in the field is unknown. Our results imply that predaceous cannibalism may have taken place four times throughout our study (one instance of nymph on adult, and three instances of adult on nymph). It seems unlikely that a third-instar nymph could actively prey upon an adult, which suggests that this may have been a case of scavenging, or of incidental mark transfer (see below for more discussion on these potential issues). Another explanation may lie in the laboratory colony origin of our L. hesperus specimens; these individuals may be more likely to exhibit cannibalism than wild-type members of the species. Regardless, this finding does suggest that a follow-up experiment using this methodology could shed more light on the question of cannibalism, or predatory behavior generally, in wild populations of L. hesperus.

A limitation of this study, and of this style of UFIT, is that intra-life stage cannibalism is impossible to detect; that is, adults preying on adults or nymphs preying on nymphs would not be detected by our method. It is also possible that what appeared to be predation events on L. hesperus nymphs and adults were actually scavenging events; as the cages were not directly observed during the 6-h period when feeding would have taken place, it could be the case that our introduced L. hesperus died and were only subsequently fed upon. While we tried to minimize this possibility through having a short window of time between introduction and collection, this scenario cannot be ruled out entirely. One final consideration which may be relevant to our results is the possibility, because of the use of external marks, for ‘near-miss’ predation attempts to transfer an ELISA-detectible dose of protein mark to the would-be predator, while the prey survives. This was found to be a concern for chewing predators, but the overall incidence is presumed to be low (Hagler et al. 2018b). It should be noted that, to our knowledge, the possibility of obtaining a ‘near-miss’ false-positive gut assay response has never been examined for by the prey-specific PCR assay approach.

In summary, this and other recent studies (see review by Hagler 2019) show that the UFIT can expedite research on identifying arthropod feeding preferences. Here, we demonstrated that the UFIT was useful for identifying stage-specific feeding events by an assemblage of predators on an economically important pest species. Data showed that S. xyloni and C. vittatus were adept at finding and feeding on the L. hesperus egg stage. Further, the whole-plant sampling technique, in combination with the UFIT gut analyses, showed that Geocoris spp. and spiders were the most abundant predator taxa and the most proficient at feeding on L. hesperus nymphs. As such, the UFIT was useful for providing more refined data on predator feeding activity than prey-specific gut assay procedures. Our methods show promise in shining light on which natural enemies put pressure on pests at the most advantageous stage for biocontrol purposes.

Disclaimer: Mention of trade names or commercial products in this publication is solely for providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity provider and employer.

Author Contributions

JRH: Conceptualization; experimental design, data curation; data analysis; resources; supervision; visualization; writing original draft. ALT: methodology; investigation, editing original draft. SAM: Data curation; investigation; methodology; data acquisition; editing original draft. MTC: Data curation; investigation; methodology; data acquisition; editing original draft.