Comparing the Effects of Spring and Fall Tillage on Larval Populations of a Beneficial Insect

Final Report for GNC12-151

Project Type: Graduate Student
Funds awarded in 2012: $9,916.00
Projected End Date: 12/31/2014
Grant Recipient: Purdue University
Region: North Central
State: Indiana
Graduate Student:
Faculty Advisor:
Dr. Ian Kaplan
Purdue University
Expand All

Project Information


Granivorous ground beetles (Coleoptera:Carabidae) are abundant and ubiquitous throughout agricultural systems, and reduce weed seedbank densities. While much research has focused on adult activity patterns and the conservation biocontrol services they provide, little is known about their biology and habitat requirements during larval stages, despite the fact that adult recruitment is determined by factors that promote larval survival. We present results of larval pitfall trap surveys of Harpalus pennsylvanicus, a common weed seed predator across North America, from two experiments examining its phenology and distribution across tillage and cover cropping gradients in organic tomato systems. Larvae emerged 4-6 weeks after the adult activity peak, and larval activity density was up to 10 times higher in no-till crop environments than in cultivated areas. Compared with adults, larvae are relatively immobile and vulnerable to disturbance; thus, weed management strategies that rely on intense cultivation may undermine the ecosystem services they provide. Growers must balance competing priorities of immediate weed suppression and future biological conrol.


Increasing demand for produce raised without chemical inputs creates an imperative for agroecologists to develop stronger, ecologically based, cultural tools for farmers to manage weeds effectively and improve crop yields. An ecological approach to weed management combines several tactics including tillage, cover cropping and conservation biological control as an alternative to a simplified herbicide program (Westerman et al. 2005). Of these tactics, tillage is most heavily relied upon in herbicide-free systems, although soil disturbance eliminates food and habitat resources for natural enemies, creating a potential trade-off between weed suppression and biocontrol. Invertebrate seed predators can substantially reduce seedbanks and affect weed population dynamics (Davis et al. 2003, Westerman et al. 2006), but suffer high mortality due to heavy tillage (Purvis and Fadl 2002, Holland and Reynolds 2003). Nonetheless, they are abundant and ubiquitous in cropping systems across varied pest management regimes and geographical regions (Bohan et al. 2011), justifying widespread application of cultural techniques to enhance seed predation services in a variety of cropping systems (Landis 2011).

Ground beetles (Coleoptera:Carabidae) are the dominant seed predator taxa in many agricultural systems, and numerous studies have tested the effects of cover and tillage systems on seed predation services by adults (Gallandt et al. 2005, Pullaro et al. 2006, Shearin et al. 2008, Meiss et al. 2010, Ward et al. 2011). Despite the wealth of data on carabid adults in cover crop systems, little is known about distributions of larvae in heterogeneous environments, largely because of their cryptic, belowground lifestyle and difficulty identifying immature stages (Lovei and Sunderland 1996). The sparse information collected on larval phenology and life history of seed-feeding carabids in crop environments is largely observational (Kirk 1972a,b,c) or lab-based (Jorgenson et al. 1997, Hartke et al. 1999, Saska 2005), whereas few experimental studies have documented larvae in field investigations (Luff 1980, Traugott 2001, Noordhuis et al. 2001 Purvis and Fadl 2002, Frank et al. 2010).

Harpalus pennsylvanicus Dej. is an important focal species, being the most common carabid seed predator in many agricultural systems across North America (Barney and Pass 1986 Davis and Liebman 2003, Lundgren et al. 2006, Ward et al. 2011, Fox et al. 2013), with a peak activity period that directly coincides with the senescence of many summer annual weeds (Kirk 1972a). Larvae (Fig 1) emerge in late autumn, and are identifiable by their enlarged heads and mandibles (Tomlin 1975), acuminate laciniae, unequal claws (Bosquet 2010), and a signature shape of the frontal margin (Kirk 1972a). They actively forage for about 4 weeks before overwintering in small burrows where they cache weed seeds (Kirk 1972c). Larvae are relatively immobile and vulnerable compared to adults, and most mortality occurs before pupation (Kromp 1999). Due to high larval mortality, adult recruitment and weed seed biological control may be largely driven by local habitat factors that promote larval success, and the effects of varying cultural treatments on larval densities warrant investigation.

Here, we report on larval activity patterns of H. pennsylvanicus within two separate field experiments. First, we compare varying cultural weed control strategies in a market tomato system including tillage, plastic mulch, living mulch and roller crimped rye mulch. Second, we compare tillage practices and variety of fall cover crop types. We hypothesized that no-till treatments, particularly those containing a killed fall cover crop would provide optimal in-crop overwintering habitat, due to enhanced larval food resources and insulating thatch. We also document predictable emergence times for H. pennsylvanicus larvae in the Midwest, and discuss cultural treatments that may enhance overwintering survival.

Project Objectives:

1) Describe the active period of H. pennsylvanicus in market vegetable systems
2) Evaluate larval activity density in varied tillage and cover crop treatment combinations
3) Compare adult and larval activity densities between several fall cover crop species


Click linked name(s) to expand
  • Dr. Ian Kaplan


Materials and methods:


We conducted both experiments at the Purdue University Meigs Horticulture Research Farm near Lafayette, IN (40° 17’ 15” N, 86° 53’ 1” W) using organic crop management practices.

 Experiment 1

Experimental design

This experiment occurred during the 2011 growing season, and compared carabid activity across four cover crop and tillage treatments in a market tomato system (Solanum lycopersicum cv. ‘Fraisers Gem’). Plots were 6x6m with 4 crop rows spaced 1.8 m apart, arranged in 5 randomized blocks, with 4.5m margins between plots and blocks. The experimental matrix had been planted with red clover (Trifolium pratense L.) over the two previous years, and was bordered on three sides by an untilled perennial grass margin. The first treatment (TILL) represented standard practices for organic fresh-market tomato production. Plots were spring-tilled and plastic mulch suppressed weeds within rows. Cultivation controlled weeds between rows, and occurred on June 8, July 6, July 27 and August 10. In the second treatment, (TILL+CLOVER), tomatoes were transplanted into plastic mulch, cultivated as needed for 8 weeks between rows, then planted with a crimson clover (Trifolium incarnatum L.) cover crop at 35 kg seeds per ha. In a third treatment (NO-TILL RYE), a cereal rye (Secale cereale L.) cover crop was drill-seeded at a rate of 135 kg seeds per ha in fall 2010 and was roller-crimped in spring 2011. Tomatoes were no-till transplanted into the rye mulch 48 hrs after crimping. The last treatment (STRIP-TILL RYE) was similar to NO-TILL RYE, but the area between rows were tilled and planted with red clover in spring 2011 (rye mulch was left within rows). Instead of cultivation, weeds were managed in the NO-TILL RYE and STRIP-TILL RYE treatments by mowing between rows on July 6, July 27, August 8, and September 8. Weed management activities were performed with a BCS 722 walking tractor in all treatments using mower and tiller attachments.


From July-November 2011, we collected carabid adults weekly from two pitfall traps filled with 1cm soapy water linked by a 75 cm aluminum flashing barrier in the center row of each plot. As temperatures fell in late autumn (the active season for H. pennsylvanicus larvae), traps were collected every two weeks, and carabid adults and larvae were deposited in vials with 70% ethanol. Pitfall traps were open constantly during the sampling periods. Heavy rain events usually caused traps in tilled plots to flood, so exact collection periods varied with inclement weather. Adults and larvae were identified using Bosquet 2010 and Kirk 1972b, and adults were confirmed with specimens in Purdue’s Entomological Research Collection (PERC). Voucher specimens were subsequently deposited in PERC. Adults were identified to species, whereas larvae were identified to the tribe level. Although most 3rd instar larvae were identified as H. pennsylvanicus, 1st and 2nd instar larvae are difficult to identify beyond tribe. Thus, all specimens were pooled in one taxonomic group within the tribe Harpalini, which was almost certainly dominated by H. pennsylvanicus. All genera within the Harpalini are opportunistically granivorous as adults (Lundgren 2009), and contribute to weed seed predation services in crop environments.

Experiment 2

Experimental Design

 In 2012, we compared four commonly utilized fall cover crops, (rye, rye/vetch, oriental mustard, and fallow) and two tillage treatments (no-till and spring tillage) in a randomized split-plot design. In fall 2011, the 4 cover crop types were planted in 4 replicate blocks of 14x14m main plots with 4.5m margins between them. The experimental matrix was fallow for 10 years previously, and surrounding margins as well as adjacent crop areas were tilled in spring 2012 and planted with sorghum-sudangrass (Sorghum × drummondii Steud). In May 2012, plots were split in half; one subplot was flail-mowed and tilled before transplanting tomatoes (cv. ‘Brandywine’). In the other subplots, cover crops were flail mowed and left as mulch on the soil surface between rows. There was no buffer nor margin between the subplot tillage treatments. Black plastic controlled weeds within rows and mowing occurred between rows as needed in all treatments. In mid-September, cultivated subplots were tilled again before fall cover crops were planted, while they were drill seeded in the no-till subplots.


Collection and identification methods were similar to Experiment 1 except that pitfall traps were not barrier linked, and had no killing agent. Adults were trapped alive, identified on-site, and released at the plot center. Larvae were collected and stored in vials with 70% ethanol. Sampling was temporarily suspended from Sept. 1-26 while tomato crops were harvested and fields were re-planted with fall cover crops. To sample larvae, we used only 4 of the 16 tilled subplots, which were structurally identical, and handled the tillage treatment as a fifth cover type for the duration of the study. Larvae have extremely limited mobility (<15cm/day; Kirk 1972a), thus the doubled pitfall trap density in the main plots were larvae were sampled was unlikely to confound our measures of larval activity density in tilled treatments. During the larval activity period (Sept. 26-Nov. 11), we also set 5 additional pitfall traps at the end of each block in an adjacent perennial margin to compare distributions in undisturbed overwintering habitat. We attempted to use multiple sampling methods for larvae throughout the season, including soil cores, quadrats, sentinel prey, and litter-bag surveys, but only pitfall traps yielded sufficient data for analyses.


We analyzed activity densities of H. pennsylvanicus adults and larvae as seasonal sums per replicate plot in both experiments. The two experiments performed in 2011 and 2012 were analyzed separately, due to differing cultural treatments, collection methods and plot sizes. We performed all analyses in R version 3.0.2 (R Core Team 2013).

 Experiment 1) Adult H. pennsylvanicus pitfall trap captures were ln(x+1) transformed and compared between the four cultural weed control treatments (TILL, TILL+CLOVER, STRIP-TILL RYE and  NO-TILL RYE) using ANOVA, and pairwise comparisons were made between treatments using Tukey’s HSD test. Because seasonal larval captures included many zero counts, they did not meet the assumptions of ANOVA, and were also inappropriate for rank-based analysis, so these data were analyzed using random permutation tests. To do this, the observed F statistic across treatments was compared to 10,000 F statistics calculated from permuted distributions of the larval dataset. P-values calculated were the proportion of randomly generated F-values that were greater than the observed F-value. Pairwise tests between treatments were made by comparing observed differences in mean larval catch for each of the 6 possible treatment combinations to 10,000 randomized mean differences from permuted distributions. Associated P-values were calculated as the proportion of randomly generated mean differences that were greater than the observed differences.

 Experiment 2) Adult H. pennsylvanicus captures were ln(x+1) transformed and cover crop treatments (fallow, mustard, rye, and rye/vetch) and the two tillage treatments were evaluated using a two-way ANOVA, with cover crop and tillage treatments as categorical predictor variables, blocked by main plot. Larval captures were also ln(x+1) transformed and the 5 (4 cover crop + 1 tilled) treatments were analyzed with a one-way ANOVA.

Research results and discussion:


Larval activity phenology

Harpalini larvae became surface-active in late October in both experiments, 4-6 weeks following the peak in adult H. pennsylvanicus activity (Fig 2a and 2b). In 2012, a peak in larval capture was observed during the 3rd week of October, but trapping was terminated too early determine an activity peak for 2011. Trap capture for adults in the 2012 experiment was extremely low, with only 20% of the mean nightly capture compared to the previous experiment, partially due to differences in trapping techniques and the presence of a killing agent in the traps in 2011.

Experiment 1)

Seasonal activity densities for H. pennsylvanicus adults were more than 5-fold greater (F=17.06, df =16, p<0.005; Fig. 3a) in the NO-TIILL RYE plots and the STRIP-TILL RYE plots with than in the cultivated treatments (TILL and TILL+CLOVER). Larval activity showed the same trends overall (significant treatment effect: F=2.92, n=20 p=0.040; Fig. 3b), but differences in seasonal activity density were only statistically significant between NO-TILL RYE and cultivated treatments (TILL and TILL+CLOVER; Fig 3b).

Experiment 2)

Unlike Experiment 1, there were no differences in the activity density of H. pennsylvanicus adults across tillage (F=0.13, df=23, p=0.714; Fig 4a) or cover crop treatments (F=1.122, df=23, p=0.361; Fig 4a). Harpalini larval activity varied across treatments (F=5.67, df=15, p=0.005),and was significantly higher in the mustard and fallow plots than in tilled plots or rye/vetch (Fig 4b), although tilled plots did not differ from rye or rye/vetch treatments.


The treatments in Experiment 1 allowed us to examine adult and larval carabid activity over an ordinal gradient of soil disturbance. Like several other studies (Brust and House 1984, Cromar et al. 1999, Shearin et al. 2008), adults foraged disproportionately in treatments with reduced cultivation (Fig 3a), but were equally active in the NO-TILL RYE and STRIP-TILL RYE treatments, even though the strip-tilled plots were moderately more disturbed. Both of those treatments had substantial weed growth (Butler 2012), which may be an important cue seed-feeders use to identify foraging environments. Larvae were almost completely absent in frequently tilled sites (Fig 3b), even when a clover cover crop was present in the TILL+CLOVER treatment. Compared with the two cultivated treatments, larval activity density was significantly higher in the least-disturbed NO-TILL RYE treatment, but not in the STRIP-TILL RYE treatment, suggesting that adult females have oviposition preferences for sites that have been free of cultivation for at least one growing season. Given that Harpalus spp. larvae have rather limited mobility on the soil surface,  (

We found additional support for carabid oviposition preference for less disturbed crop environments in Experiment 2. Although larvae have been observed before in cultivated fields (Kirk 1972a), our lowest larval captures were in tilled plots. One other study (Brust and House 1990) reported higher carabid larval densities and higher corn rootworm predation rates in no-till environments during late summer, but larvae were only identified at the family level, which prevents inferences about granivore feeding guilds. H. pensylvanicus overwinters in the vulnerable larval stage, and may be limited by thatch-insulated overwintering habitat and dispersal ability (Hof and Bright 2010, Fox et al. 2013). With this in mind, it was surprising that we observed lower activity densities in the rye and rye/vetch plots (Fig 4b), as those treatments were the only ones with undecomposed cover crop residue remaining in late autumn. The denser thatch layer in the rye and rye/vetch plots may have been a structural impediment for foraging larvae, reducing their likelihood of capture in those treatments. Alternatively, live weed biomass was highest between rows in the mustard and fallow plots (data will be reported elsewhere), and may provide superior food and shelter resources to overwintering larvae.

We attempted to compare larval activity in perennial habitats with our experimental treatments by sampling the nearest un-tilled margin during the activity peak in 2012, but captured very few larvae (data not shown). Again, it is likely that the grassy thatch layer in perennial margins impedes movement, and pitfall trap captures may be a poor estimate of foraging activity in heavy soil-surface vegetation (Thomas et al. 2006). Studies evaluating directional movement between crop and perennial non-crop habitat have shown clear seasonal dispersal patterns by seed-feeding carabids (Best et al. 1981, Thomas et al. 1997), and undisturbed non-crop habitats are unequivocally critical as refuge for overwintering natural enemies (Tillman et al. 2012). Improved sampling strategies must be employed to more accurately describe spatial distributions of carabid larvae in heterogeneous habitats, particularly those species which are less surface active as larvae.

Unlike Experiment 1, we found no significant differences in activity density of H. pennsylvanicus adults between cover crop types, nor between tillage treatments in Experiment 2 (Fig 4a).The cultivated treatments differed from Experiment 1 in that cultivation occurred only twice; once in the spring shortly before crop planting, and again in September before fall cover crops were re-planted. In the 5 months between cultivation events, weeds were mowed between crop rows. Ward et al. (2011) found a similar absence of effects due to infrequent tillage with H. pennsylvanicus foraging in sweet corn. Sparse weed cover and longer disturbance intervals may provide adult beetles with adequate foraging habitat (Shearin et al. 2007), but requirements may be more stringent when selecting safe overwintering and oviposition sites.

In both experiments, we found that activity densities of larvae were not consistent with habitat use by adults; adults tolerated infrequent (annual) tillage (Figs 3a, and 4a), but larvae appear sensitive to soil disturbance, even when it occurs as much as six months prior (Fig 3b). This habitat-use discrepancy is important because nearly all the work done to quantify carabid communities and their ecosystem services in crop environments focuses only on adults; this may not accurately describe the biotic costs of cultivation practices for carabid communities. Across all life stages, carabids common in agricultural systems may be less resistant to seasonal disturbance than assumed (Holland and Luff 2000, Jonason et al. 2013).

 Recommendations for enhancing weed seed predation services on-farm are to delay tillage until late in the fall or spring (Menalled 2008, Ward et al. 2008). This strategy maximizes the time weed seeds are exposed to soil-surface seed predators, but does not consider oviposition preference of invertebrate granivores. Delaying tillage may provide an optimal egg laying site due to enhanced weed cover and food resources (Lovei and Sunderland 1996) in the fall when many carabids breed, but a substantial population of larvae may be compromised if tillage occurs the following spring (Purvis and Fadl 2002). This study demonstrates that larval activity densities are disproportionately high in sites without fall cultivation, and further research must examine source/sink relationships between seed-predator recruitment and tillage timing.


Participation Summary

Educational & Outreach Activities

Participation Summary

Education/outreach description:


Blubaugh, C.K. and I. Kaplan. 2014. Weed management tactics compromise recruitment of weed seed predators across developmental stages. Prepared for submission to BioControl

Non-peer reviewed outreach publications:

Blubaugh, C.K. 2013. Organic Insect Pest Management. Purdue University College of Agrigulture. <>

Blubaugh, C.K. 2012. Research update: Cover crops and weed seed predation services. Purdue Vegetable Crops Hotline. 561:3-4.

Blubaugh, C.K. 2012. Meet the (beneficial) beetles: Harpalus pennsylvanicus, weed seed predator. Purdue Vegetable Crops Hotline. 559:2.

Outreach Presentations:

Indiana Small Farm Conference Danville, IN (2/2014)

“Cultivation-based weed management tactics compromise recruitment of weed seed predators across developmental stages.” (research talk)

Meigs Research Farm Tour, Lafayette, IN (8/2013)
“Cover crops and intra-guild predators weed mediate weed seed predation services” (demonstration at the research site)

Indiana Small Farm Conference Danville, IN (3/2013)

“Seed predators reduce weed germination in cover crops” (poster)

 Great Lakes Vegetable Working Group: Annual meeting West Lafayette, IN (2/2013)

“Cover crops and baby carabids: how does larval environment shape a beneficial insect community?” (research talk)

 MOSES Organic Farming Conference LaCrosse, WI (2/2013)

“Seed predators reduce weed germination in cover crops” (poster)

 Local Growers Guild Webinar (1/2013)

“A brief tour of new organic research at Purdue” (Viewed online by growers throughout IN)

Indiana Vegetable Growers Association (1/2013)
“Cover crops and baby carabids: how does larval environment shape a beneficial insect community?” (research talk)

Project Outcomes

Project outcomes:

  • Our results emphasize the importance of examining weed seed predators across all life stages.  Habitat requirements vary beween adults and larvae, and managing beneficial insect habitat based on the activity and episodic seed-removal estimates may compromise immature stages, and thus future weed seed predation services.

  • Activity densities of Harpalini larvae were higher in plots where tillage had not occurred for a full growing season. Incorporating a fallow season or full-year cover crop in a farm management plan may improve the source pool of seed-feeding carabids to colonize adjacent areas in subsequent growing seasons.

  • Activity densities of Harpalini larvae and adults were also much higher in sites with weed growth. Tolerating benign weed growth (growth that neither threatens crop yield nor contributes to the seedbank) may improve overwintering habitats for larvae.

Farmer Adoption

We engaged growers at several outreach events, and definitely improved consciousness of the ecosystem services weed seed predators provide. All growers make complicated weed management decisions where they must consider competing priorities of yield protection, labor investment, seedbank flux and biological control. This work along with several other projects focused on the farm-scale impacts of weed seed predation helps growers adopt a comprehensive, whole-farm approach to seedbank management, advocating for the maintenance of environments where crops and natural enemies both flourish.


Areas needing additional study

  • This study focused on Harpalus pennsylvanicus and other larvae in the tribe Harpalini.  While this is likely the most common seed predator group of carabids, other species within the tribes Zabrini and Pterostichini also perform weed seed predation services. Due to the difficulty of sampling larvae from these two groups, little is known about their biology and life history in immature stages.  Future research could describe the phenology and habitat requirements of larvae from other taxonomic groups in North America.

  • Our data do not reflect absolute densities of carabid larvae in our system, a major limitation of using pitfall traps to survey insects (Greenslade 1964). This method precludes sampling of carabid larvae that are exclusively subterannean, and relatively few carabid species are surface-active.  We made substantial efforts at other sampling methods that estimate absolute densities of larvae, but had great difficulty recovering sufficiant data.  Future studies could employ intensive soil survey methods or emergence traps (Holland et al 2003) to estimate absolute densities of larvae in the field.

  • Finally, future research can address trade-offs between cultivation and future biological control services.  We showed that delayed-tillage creates an ideal environment for larvae overwintering, but little is known about the fates of larvae in subsequent seasons. Fencing plots with varying cultivation treatments combined with seed predator sampling over multiple growing seasons would yield interesting information about how disturbance affects future generations of biological control agents. Evaluating seed predation and weed recruitment in the same framework could lead to further knowledge about the lasting effects of cultivation on biocontrol.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.