SEARCH

SEARCH BY CITATION

Keywords:

  • direct density dependence;
  • inverse density dependence;
  • invertebrate seed predators;
  • vertebrate seed predators;
  • weed population dynamics

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Seed predation can cause substantial losses of newly produced weed seeds and can therefore be important in regulating weed densities. The impact on weed population dynamics is greatest if predation acts in a directly density-dependent manner.
  • 2
    We investigated the effect of between-patch variability in seed density on seed removal. For this purpose, artificial weed seed patches were created by broadcasting giant foxtail Setaria faberi, at low (1000 seeds m−2), medium (4500 seeds m−2) and high densities (9500 seeds m−2) over 25 ¥ 25 m areas within three maize fields in August. Changes in giant foxtail seed densities were evaluated 3 weeks and 7 weeks post-application, using soil surface sampling.
  • 3
    Observations of seed predation rate (seeds seed−1 week−1), using seed cards and exclosure cages, activity–densities of invertebrates using pitfall traps, and population estimates of rodents using Sherman live traps, were conducted to understand and explain the dynamics of seeds on the soil surface.
  • 4
    Three weeks after seed addition, seed predation was strongly and inversely density dependent. After 7 weeks, the net response since the start of the experiment exhibited only a weak inverse density dependence. This means that between 3 weeks and 7 weeks after seed addition, the response had reversed from inverse to almost direct density dependence.
  • 5
    During the August–October period, seeds in maize fields were mainly consumed by invertebrates. The most abundant granivorous invertebrates were crickets Gryllus pennsylvanicus and Allonemobius allardi, and carabid beetles, especially Harpalus pensylvanicus. The insects appeared unable to detect and respond numerically to weed patches, resulting in inversely density-dependent predation, which favours the persistence of weed patches. The granivorous prairie deer mouse Peromyscus maniculatus bairdi, was present but contributed little to overall seed losses in autumn.
  • 6
    Synthesis and applications. The results of this study indicate that weed densities in maize fields currently are not regulated through directly density-dependent seed predation, because the time between seed shed and seed movement into soil is too short for invertebrates to respond to and level out spatial differences at the scale of weed patches. However, our results suggest that delaying crop harvest and tillage may provide invertebrate predators with more time to attack weed seeds, and may allow for subsequent predation by vertebrates, which would be directly density-dependent.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Modelling studies have shown that seed mortality is among the most influential life cycle parameters of weeds (González-Andujar & Fernandez-Quintanilla 1991; Jordan et al. 1995; Davis, Dixon & Liebman 2003). Seed predation can cause substantial losses of newly produced weed seeds (Harrison, Regnier & Schmoll 2003; Westerman et al. 2003), and can therefore be important in regulating weed densities, especially if seeds are consumed in a directly density-dependent manner (Turchin 1995; Neubert & Caswell 2000). Weeds usually occur in patches, which tend to persist over time and serve as a source for further weed dispersal (Cousens et al. 2006). Density-dependent seed predation at the scale of weed patches could eliminate weed patches or at least limit their persistence and retard their spread.

So far, studies of momentary predation rates in agro-ecosystems have been inconclusive or contradictory with regard to the response of seed predators to seed density. For example, Brust & House (1988) found a density-independent response to seeds of four broadleaf weeds in soyabean fields; Cardina et al. (1996) found an inversely density-dependent response to Abutilon theophrasti Medicus seed densities in maize fields; and Cromar, Murphy & Swanton (1999) reported a directly density-dependent response to seeds of Chenopodium album L. and Echinocloa crus-galli (L.) Beauv. in maize.

A possible explanation for the variability in the outcome of the above studies is the involvement of different or multiple seed predators, which operate on different spatial and temporal scales and which may differ in functional and numerical responses to seed densities (Hulme 1997; Marino et al. 2005). Heads & Lawton (1983) predicted that for a specific predator, predation would change from inverse to direct density dependence with increasing size of patches. Patches beyond the action radius of the predator are too large to be treated as a patch. By analogy, we postulate that for a specific patch size the response will change from inverse to direct density dependence with increasing mobility of the predator. Birds are very mobile and can differentiate between fields that differ in food availability (Moorcroft et al. 2002). Rodents can differentiate between patches within fields (Angelstam, Hansson & Pehrsson 1987; Tew, Todd & Macdonald 2000), and invertebrates between small patches and individual weed plants (Zhang et al. 1997a; Honěk, Martinková & Saska 2005). A density-dependent response, if any, will therefore be apparent at different spatial scales, depending on the action radius of the predator.

Arable fields in Iowa, USA, harbour a diverse set of seed predators, including crickets, Gryllus pennsylvanicus Burmeister and Allonemobius allardi (Alexander and Thomas), carabid beetles, especially Harpalus pensylvanicus DeGeer, and rodents, particularly Peromyscus maniculatus bairdi (Wagner) (Heggenstaller et al. 2006; O’Rourke et al. 2006). All are known to consume weed seeds, including giant foxtail Setaria faberi Herrm. (crickets and H. pensylvanicus: Lund & Turpin 1977; Barney & Pass 1986; Carmona, Menalled & Landis 1999; P. maniculatus: Clark & Young 1986; Getz & Brighty 1986; J. Andjelkovic and B. Danielson, unpublished data), but none is exclusively granivorous. Their diet is supplemented with both dead and living insects, or green and dead plant matter [crickets: Criddle 1925 (cited in Byers & Barratt 1991); Carmona et al. 1999; H. pensylvanicus: Best & Beegle 1977; Barney & Pass 1986; P. maniculatus: Holling 1959; Parmenter & MacMahon 1988]. However, rodents and carabid larvae also need to cache seeds for winter storage (Kirk 1972; Zhang et al. 1997b).

In this study, we investigated whether density-dependent predation of weed seeds could play a role in regulating densities of annual weeds in agro-ecosystems. For this purpose, seeds of giant foxtail were broadcast at three densities (1000, 4500 and 9500 seeds m−2) over designated subplots (25 × 25 m) within three maize fields in Iowa, USA. Changes in giant foxtail seed densities were followed over time. The range of seed densities applied reflects the range of densities naturally encountered in arable fields (Leguizamón & Roberts 1982; Forcella, Peterson & Barbour 1996; Westerman et al. 2003), while subplot size approximates the average size of weed patches (Cousens et al. 2006). The observation period started in August when weed seeds are normally shed in Iowa maize fields and ended in October with maize harvest. Crop harvest, which is accompanied by the accumulation of litter on the soil surface, results in seeds at the soil surface being covered, hidden, and potentially out of reach of seed predators (Westerman et al. 2006). Understanding seed predation dynamics, in terms of predator identity, predation rates and predator numerical response, was a second objective, and therefore changes in instantaneous predation rates and the relative distributions of predator population over subplots were followed over time.

We assumed density independence as a null hypothesis, i.e. a range of seed densities is provided: over this range we assume a constant proportion predation, although we realize that the size of the subplots was such that it might elicit a directly density-dependent response from the vertebrates and a density-independent or inversely density-dependent response from the less mobile invertebrates.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Trials were conducted from August to October 2006 in three maize fields managed by Iowa State University near Ames, Iowa, USA. The fields selected for the trial (Agronomy Farm, 5·3 ha; Bennett Farm, 86·0 ha; Kelley Farm, 6·5 ha) were located 5–8 km apart. A description of crop management practices is provided in Appendix S1, Supplementary material.

experimental design

In each field, a 50 × 50 m plot was established at a distance of at least 50 m from the nearest field edge. Plots were divided into four equal 25 × 25 m subplots. Giant foxtail seeds were added to the subplots at 0 (Control), 1000 (Low), 4500 (Medium) or 9500 (High) seeds m−2. The seed density treatments were randomly assigned to each of the four subplots per field. Seeds were manually applied at Agronomy Farm on 14 August 2006, at Bennett Farm on 15 August 2006, and at Kelley Farm on 16 August 2006. To ensure a uniform coverage of the subplot areas, pre-weighed amounts of seeds were applied to each maize row (0·76 m between-row distance × 25 m length = 19 m2).

Setaria faberi Herrm. (Poaceae), giant foxtail or Japanese bristlegrass, was used because of its demonstrated palatability to vertebrate and invertebrate seed predators (Heggenstaller et al. 2006; O’Rourke et al. 2006). It was introduced from China into the USA in the 1920s, and is now common in maize fields in the central Corn Belt. A large batch of giant foxtail seeds was gravity and colour separated by the Iowa State University Seed Science Center, Ames, Iowa. Seed viability was 57% (± 3·4% SE), determined from eight samples of 25 seeds taken randomly from the bulk seed.

Seed densities available to seed predators deviated from the seed densities applied because of low natural infestations of giant foxtail and due to germination. We assumed that germinated seeds were unsuitable as food to the seed predators, although there is evidence that seed predators might consume young seedlings under certain circumstances (Clark & Young 1986; Byers & Barratt 1991; Zhang et al. 1997b). However, this probably does not happen if large numbers of seeds are available. The initial amount of seeds available to predators (i) was corrected via:

  • i = (p + a) × (1 – g) (seeds m−2)(eqn 1)

with p, the existing density of giant foxtail seeds before seed addition; a, the density of giant foxtail seeds applied; and g, the proportion germinated giant foxtail seeds. Pre-existing weed seed densities, p, were determined by sampling the surface soil in each subplot on 8 August 2006, a week prior to seed additions. Further details are provided in Appendix S2, Supplementary material and the results are summarized in Table S1, Supplementary Material.

A 5 × 5 sampling grid, with points spaced 3·25 m apart, was generated within the central 13 × 13 m area of each subplot. The 25 sampling points were used to obtain information on seed predation rates and predator activity–densities (see below).

relationship between initial seed density and seed removal

Changes in giant foxtail seed densities, S, over time were estimated by post-application soil sampling on 5 September and 2 October 2006. Surface soil was collected to a depth of approximately 1–2 cm in 10 randomly selected 0·1 m2 quadrats within each 25 × 25 m subplot. Seeds were separated from soil using an elutriator (Wiles et al. 1996) and an air column separator. The proportion of seeds lost, Rt,t+1 = (St – St+1)/St, was calculated from the average number of seeds per square metre, St, and St+1, in two subsequent sampling periods. Similarly, the proportion of seeds lost over the entire period (15 August–2 October) was calculated as Rt,t+2 = (St – St+2)/St. Control plots were excluded from this analysis. R was regressed on St using linear regression analysis (genstat GLM, Poisson distribution, log-link function) with farm (Agronomy, Bennett or Kelley) as an explanatory factor. The null hypothesis, i.e. density-independent seed predation, was rejected if the slope of the regression line was significantly different from 0 (Forrester & Steele 2000). Although this method has been disputed (see for review Turchin 1995), it is currently the standard for detecting density dependence.

We recognize that seed densities estimated by the above soil sampling procedure may not represent the real seed densities available to seed predators. We sampled to a depth of 1–2 cm, whereas invertebrates will mainly forage on the soil surface, and, although rodents can recover buried seeds, the recovery rate is reduced compared to surface seeds (Hulme 1994).

seed predation rates

Seed cards containing 50 seeds of S. faberi were exposed to predators for 7-day periods, immediately following seed addition (Agronomy Farm, 14–21 August; Bennett Farm, 15–22 August; Kelley Farm, 16–23 August), following the intermediate soil sampling (13–21 September and 20–27 September), and following final soil sampling (6–13 October). Seed cards were constructed as described by Westerman et al. 2003. Cards were exposed in the field (i) within a cage that excluded vertebrates but not invertebrates (six per subplot; Invert); (ii) without a cage and thus accessible to both vertebrate and invertebrate predators (six per subplot; Open); or (iii) within a cage that excluded both vertebrates and invertebrates (six per subplot; Control). Exclosure cages (12 × 12 × 5 cm) were constructed from 10 mm-mesh (Invert) or 3 mm-mesh metal screen (Control) and secured to the soil using steel pegs. Locations and exposure treatments were randomly assigned to 18 of the 20 available sampling points; five sampling points were avoided because they were occupied by pitfall traps (see below).

After retrieval, the numbers of seeds remaining on Control (Ncontrol), Invert (Ninvert), and Open cards (Nopen) were counted. The effects of location (Agronomy Farm, Bennett Farm, Kelley Farm), exposure treatment (Open, Invert, Control), and applied seed density (Control, Low, Medium, High) on the number of removed seeds per card were tested for each 7-day exposure period. Generalised linear mixed models (procedure IRREML, binomial distribution, logit-link) were fitted (Genstat 5 Committee 1993). Although the 7-day exposure period in August was shifted by a day between farms, the data were treated as one data set. Results are presented as the proportions of seeds lost due to either invertebrates or vertebrates relative to the number of seeds that remained on the control cards (Abbott 1945): Minvert = (NcontrolNinvert)/Ncontrol and Mvert. = (NinvertNopen)/Ncontrol (seeds consumed × seeds exposed−1 7 day−1).

identity of seed predators, relative activity–density of insects, and rodent population estimates

Insects and rodents were trapped simultaneously in all three fields for three consecutive days and nights during four sampling periods [1–4 August (pre-application), 20–23 August, 18–21 September, 15–18 October]. Sampling periods occurred during new moon phases when the movement of rodents was least impaired by moonlight (Plesner Jensen & Honess 1995).

Invertebrate seed predators were monitored using pitfall traps that were randomly assigned to five of the 16 sampling points per subplot that were not occupied by Sherman traps. Pitfall traps consisted of 950 mL containers (11 cm diameter; 13 cm high), buried flush to the soil surface, and filled with 150 mL of 20% propylene glycol solution. They remained closed in the field until use. Granivorous insects were identified to species and summed per subplot and per trapping session as a measure of the activity–density of invertebrate seed predators. Factors such as temperature, humidity, beetle size, hunger level, and vegetation cover, can affect insect mobility rather than abundance (Sunderland et al. 1995), and therefore, only the relative activity densities were compared among trapping sessions.

Rodents were monitored using Sherman traps set up at nine of the 25 sampling points in a fixed 3 × 3 grid, spaced 6·5 m apart. Captured rodents were weighed, sexed, ear-tagged, and released. The proportion recapture always exceeded 50% after three nights. Given the subplot size it is more likely that a response to seed density will result from a change in habitat use, i.e. resident rodents spending more foraging time in high density plots, than from an increase in rodent abundance. For this reason, we used the total number of rodent captures per subplot and per trapping period, i.e. per 27 trap–nights (9 traps × 3 nights) as a measure of habitat use.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

relationship between initial seed density and seed removal

The proportions of seeds lost did not differ between farms, and therefore, data were pooled for all three data sets [P = 0·23 (15 August–5 September), P = 0·38 (15 August–2 October), and P = 0·75 (5 September–2 October)]. Between the seed additions on 15 August and soil sampling on 5 September (a period of 3 weeks), the slope, b1, of the regression line of the proportion of giant foxtail seeds lost from the subplots, R, on the initial seed density was significantly less than 0 (b1 = –7·7 × 10−5, P = 0·013, and the percentage accounted for by the regression line, inline image). This indicates an inversely density-dependent response (Fig. 1a). Between seed addition on 15 August and soil sampling on 2 October (7-week period), the slope of the regression line on the initial seed density was still significantly less than 0 but the regression was barely significant (b1 = –2·2 × 10−5, P = 0·046, inline image).Thus, although the response was still inversely density-dependent, the strong effect observed during the first 3 weeks of the experiment was largely dissipated (Fig. 1b). Apparently, between 5 September and 2 October, the predation rate response to seed density had reversed from a decreasing to an increasing R with increasing seed density, as shown in Fig. 1c. The slope of the latter regression line was not significantly different from 0 (b1 = –2·3 × 10−5, P = 0·065, inline image).

image

Figure 1. Relationship between the density of Setaria faberi seeds on the first evaluation date (15 August 2006), and the proportion Setaria faberi seeds lost between the first and second evaluation date (15 August–5 September 2006) (a), and between the first and third evaluation date (15 August–2 October 2006) (b); and the relationship between the density of Setaria faberi seeds on the second evaluation date (5 September 2006) and the proportion Setaria faberi seeds lost between the second and the third evaluation date (5 September–2 October 2006) (c). Data points refer to Agronomy Farm (inline image), Bennett Farm (inline image), and Kelley Farm (inline image).

Download figure to PowerPoint

Two data points were negative due to the fact that the average number of seeds recovered per square metre on a sampling date was lower than on the preceding sampling date, which was probably caused by a relatively large sampling error. One of these points, the Low density subplot at Agronomy Farm, was discarded, because the average number of seeds recovered per square metre had been low on both sampling occasions (5 September, 133 seeds m−2; 2 October, 263 seeds m−2), causing a relatively large error and an unrealistic value for the proportion predation (R = –0·98). The relative error of the second negative point, the High density subplot at Kelley farm, was much smaller (R = –0·16) and the point was therefore retained.

seed predation

All explanatory variables (location, seed density, exposure treatment and their interactions) contributed to explaining the number of recovered seeds per card during one or more of the 7-day observation periods (Table 1). The effect of the exposure treatment (Control, Invert or Vert) was always highly significant and the numbers of seeds left on the Control cards were higher than on treatment cards, indicating that seed losses from the cards were caused by predation.

Table 1.  The effects of location (Agronomy Farm, Bennett Farm, or Kelley Farm), applied seed density (low, medium, or high), exposure treatment (control, invert, or vert) and interaction terms on the number of Setaria faberi seeds recovered after 7-day exposure periods (P value)
EffectExposure period
15–22 August*13–21 September20–27 September6–13 October
  • *

    Shifted observation period: Agronomy Farm, 14–21 August; Bennett Farm, 15–22 August; Kelley Farm, 16–23 August.

Location< 0·0010·1470·003< 0·001
Seed density0·0250·116< 0·0010·151
Exposure< 0·001< 0·001< 0·001< 0·001
Location × seed density0·0240·8480·0020·236
Location × exposure0·0210·024< 0·001< 0·001
Seed density × exposure< 0·001< 0·0010·0720·749
Location × seed density × exposure0·7150·0060·0290·416

Predation by invertebrates was substantial on all three farms (Fig. 2). On Agronomy Farm, invertebrate predation peaked on 13 October (Fig. 2a) while on Bennett and Kelley Farms it peaked on 21 September (Fig. 2b,c). Immediately after seed addition, invertebrate seed predation was high in the Control subplots and low in all other subplots, on all three farms alike. This suggests that the pulse of giant foxtail seeds temporarily satiated the resident invertebrate predators. After the initial response, the trend in invertebrate predation was rather unpredictable and seemed unrelated to weed seed densities in the subplots.

image

Figure 2. Seed predation rate (week−1) as a function of time: the proportion Setaria faberi seeds removed from seed cards by invertebrates (a, b, c) or vertebrates (d, e, f), at Agronomy Farm (a, d), Bennett Farm (b, e) and Kelley Farm (c, f), in Control plots (− − −), or plots with Low (1000 seeds m−2; - - -), Medium (4500 seeds m−2; – - – - –) and High (9500 seeds m−2; —) Setaria faberi seed densities. Points refer to the day the seed cards were retrieved from the field. Arrows indicate the moment of seed addition (15 August) and the evaluation dates (5 September and 2 October).

Download figure to PowerPoint

Predation by vertebrates was important on Agronomy Farm only and peaked on 27 September (Fig. 2d). Vertebrate predation on Agronomy Farm was highest in the High density subplot, intermediate in the Medium and Control subplots, and low in the Low density subplot. On Bennett and Kelley Farms, vertebrate predation rate fluctuated between 0 and 0·4 week−1 with no apparent maximum (Fig. 2e,f). The trends in vertebrate predation on the latter two farms seem to be unrelated to weed seed densities in the plots.

relative activity–density of insects

Granivorous insects, Gryllus pennsylvanicus, Allonemobius allardi, Harpalus pensylvanicus and Harpalus herbivegus, constituted 88% of the catch of the pitfall traps at Agronomy Farm, 71% at Bennett Farm, and 95% at Kelley Farm (Table 2). Carnivorous insects caught were Poecilus chalcites (Say), P. lucublandus (Say), Agonum placidum (Say) and Cyclotrachelus sodalis LeConte. Carabid larvae constituted 21% of the catch at Bennett Farm and, although we did not identify the larvae, we assumed most were H. pensylvanicus, given the high density of adults of this species. Presumably, overnight frost (–3 °C) on 12–15 October 2006 was the reason for one or no insects being caught during the last trapping session at Kelley and Agronomy Farms, respectively.

Table 2.  Total insect catches from 20 traps per subplot over 12 days of trapping on Agronomy, Bennett, and Kelley Farms
 Agronomy FarmBennett FarmKelley Farm
No. of insects%No. of insects%No. of insects%
Orthoptera, Gryllidae
 Gryllus pennsylvanicus6643 772015251
 Allonemobius allardi 7 5 31 811237
Coleoptera, Carabidae
 Harpalus pensylvanicus6140 8622 20 7
 Harpalus herbivegus 0 0  2 1  0 0
 Poecilus chalcites 4 3  4 1  4 1
 Poecilus lucublandus 3 2  3 1  7 2
 Cyclotrachelus sodalist10 710828  4 1
 Agonum placidum 2 1  0 0  1 0
 Carabidae larvae 0 0 8121  0 0

The relative activity–densities of granivorous insects in the four subplots changed over time (Fig. 3). At Agronomy Farm, the highest number of granivorous insects was found in the Low (4 August and 21 September) or Medium (23 August) subplots (Fig. 3a). At Bennett Farm, the relative activity–densities started off and remained highest in the Medium and High subplots (Fig. 3b). At Kelley Farm, the relative activity–density seemed to change over time in favour of the Medium and High seed density subplots (Fig. 3c). Overall, there was little or no correlation between the distribution of insect numbers over subplots and weed seed densities in those subplots.

image

Figure 3. Relative activity–density of granivorous insects, caught per five traps over 3 days during four trapping sessions in autumn 2006, over Control inline image, Low inline image, Medium inline image, and High inline imageSetaria faberi seed density subplots, at Agronomy Farm (a), Bennett Farm (b) and Kelley Farm (c). The first trapping session was conducted prior to seed addition. Numbers above each column pertain to the number of granivorous insects caught during a session. Dates refer to the day the pitfall traps were emptied.

Download figure to PowerPoint

rodent population estimate and habitat use

The prairie deer mouse Peromyscus maniculatus, dominated the rodent communities, along with lower numbers of western harvest mice Reithrodontomys megalotis (Baird) (Agronomy Farm, 2), house mice Mus musculus Linnaeus (Agronomy Farm, 2; Bennett Farm, 1) and northern grasshopper mice Onychomys leucogaster breviauritus Hollister (Agronomy Farm, 3). The latter species is mostly carnivorous and was therefore omitted from counts. The total number of rodent captures during 27 trap–nights changed over time (Fig. 4), but appeared unrelated to weed seed densities, insect densities, or vertebrate predation rates in subplots (Fig. 2).

image

Figure 4. The total number of rodent captures per subplot per 27 trap–nights in four trapping sessions in autumn 2006, over Control inline image, Low inline image, Medium inline image, and High inline image Setaria faberi seed density subplots, at Agronomy Farm (a), Bennett Farm (b) and Kelley Farm (c). Numbers above each column pertain to the number of rodents (excluding Onychomys leucogaster) caught during a session. The first trapping session was conducted prior to seed addition. Dates refer to the last trapping day of each 3-night trapping session.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We studied seed predation in relation to weed seed density on the premise that a directly density-dependent relationship would help limit weed population growth by reducing densities of seeds on the soil surface in weed patches. The existence of an inversely density-dependent relationship as depicted in Fig. 1a,b was unexpected and may represent one of the mechanisms by which weed patches persist spatially. It could actually destabilize weed population dynamics by favouring weed growth in areas with high weed densities (Turchin 1995; Neubert & Caswell 2000).

The predation rates as measured by seed cards indicated that most seeds were consumed by invertebrates rather than vertebrates, and that invertebrate seed predation in the seed-supplemented subplots increased over time. Rodent seed predation rates were substantial at Agronomy Farm only. Although Peromyscus maniculatus is an important weed seed predator in the US Corn Belt, it is also an opportunistic omnivore whose diet includes insects. It is possible that P. maniculatus was actually feeding on the abundant crickets and carabids rather than on weed seeds. The response to differences in seed availability may be faster than reported here, because both granivorous insects and rodents are known to aggregate in weedy patches in crop fields (Angelstam et al. 1987; Bosch 1987; Tew et al. 2000; Hough-Goldstein, VanGessel & Wilson 2004), enabling them to be at locations prior to and during weed seed shed. Consequently, we may have underestimated direct density dependence.

There was little correlation between insect activity–density and the ranking of invertebrate seed predation rates among subplots. Neither was there convincing evidence of insect aggregation in high density plots. The low mobility combined with the omnivorous feeding habit of the insects may have prevented a numerical response to the seed densities provided. Alternatively, the data from pitfall traps may have been misleading as they are plagued by problems of interpretation related to factors that affect insect mobility (Sunderland et al. 1995).

Similarly, there was no correlation between rodent captures and vertebrate seed predation rates. A similar lack of correspondence has been reported by others (Hulme 1994; Manson & Stiles 1998; Kollmann & Bassin 2001; Reed, Kaufman & Kaufman 2004). Neither was there convincing evidence of a larger allocation of time spent in high density plots. Either the rodents did not spend more time foraging in areas of higher seed density, or the data from the mark–recapture trapping were misleading. It is known that live-trapping is not the most suitable technique for determining micro-habitat selection and resource utilization by rodents (Price 1977). The quantification of foraging movements via, for example, radio tracking or fluorescent trail analysis have been suggested as more appropriate alternatives (Douglass 1989; McShea & Gilles 1992).

Ignoring the information on predator numbers, the following picture emerges, in which crickets and carabid beetles were mainly responsible for predation on weed seeds in Iowa maize fields in late summer and early autumn. The invertebrates appeared unable to respond quickly to areas of high seed density at the 25 × 25 m spatial scale, and therefore responded initially in an inversely density-dependent manner, with high predation rates in the areas with the low seed densities and vice versa. With time, either an increasing number of crickets and carabid beetles aggregated in the high density plots, or the invertebrates present switched to a more granivorous diet, causing a directly density-dependent response, with high predation rates in the areas with the high seed density and vice versa.

Our results suggest that invertebrates may not be the most optimal predator for dealing with seeds that are temporarily available in large patches; mobility of the invertebrate predators was too low to quickly detect and respond numerically to differences in prey availability at the scale investigated here. Furthermore, invertebrates may have limited ability to respond functionally to high prey availability due to their small size and temperature-dependent activity levels. However, invertebrates may be more suitable to deal with within-patch variability, which was not investigated here.

Rodents took little or no interest in the seeds provided, except on Agronomy Farm. Instead, they may have been feeding on the available crickets and carabid beetles, or other food items. It would be particularly interesting to learn what would happen in terms of seed predation later during the season when low temperatures force carabid beetles into hibernation and kill the remaining crickets. Invertebrate seed predation would come to a stop and, at the same time, invertebrates would no longer be available as a food source to the rodents. It is likely that rodents would resort to consuming and caching (weed) seeds, given that we know that seed predation by vertebrates continues through winter and spring (P.R. Westerman, M. Liebman, personal observations). Although, we currently have little information on factors that influence rodent activity on arable fields in winter, we do know that they respond to disturbances (Cardina et al. 1996) and to the presence of cover (Heggenstaller et al. 2006).

The use of patches that are realistically large and that contain seed densities that are realistically high appeared to be essential in revealing a predator response with practical relevance. Estimates of density-dependent demographic parameters are dependent on the spatial scale employed (Heads & Lawton 1983; Marino et al. 2005), which highlights the necessity of working at relevant spatial scales. Although the results obtained from studies that use small feeding stations and lower seed densities may be relevant for understanding density-dependent processes occurring at the within-patch or between-plant level, they are probably irrelevant at the weed-patch level.

Our study has implications for the effectiveness of weed control through weed seed predation. Measures that prolong weed seed exposure on the soil surface, in particular delaying crop harvest and tillage, would allow invertebrate seed predators more time to level out differences in seed density and may facilitate subsequent predation by rodents. Given the higher mobility of rodents, these patterns of seed predation are likely to be directly density-dependent (Hulme 1993; Hulme & Hunt 1999; Ruscoe et al. 2005). The timing of harvest may therefore be decisive for the final shape of the density-dependent relationship and thus for the effectiveness of weed regulation.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We gratefully acknowledge the field assistance of D. Thompson, M. Simic, T. Lux and D. Sundberg. We would like to thank A. Gaul and E. Christian of the ISU Seed Science Center for locating, providing and cleaning a huge amount of S. faberi seeds. We thank the ISU farm managers M. Fiscus and D. Starrett, Iowa State Research Farms (ISRF) superintendent K. Berns, and Research Farms coordinator M. Honeyman for their hospitality, service and patience. Furthermore, we would like to thank three anonymous reviewers for their comments and suggestions for improvement. Financial support was provided by the USDA National Research Initiative (Project 2006-35320-16548).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Abbott, W.S. (1945) A method of computing the effectiveness of an insecticide. Journal of Economic Entomology, 18, 265267.
  • Angelstam, P., Hansson, L. & Pehrsson, S. (1987) Distribution borders of field mice Apodemus: importance of seed abundance and landscape composition. Oikos, 50, 123130.
  • Barney, R.J. & Pass, B.C. (1986) Foraging behaviour and feeding preference of ground beetles (Coleoptera: Carabidae) in Kentucky alfalfa. Journal of Economic Entomology, 79, 13341337.
  • Best, R.L. & Beegle, C.C. (1977) Food preference of five species of carabids commonly found in Iowa cornfields. Environmental Entomology, 6, 912.
  • Bosch, J. (1987) The influence of some dominating weeds on beneficial arthropods and pests in a sugar beet field. Journal of Plant Diseases and Protection, 94, 398408.
  • Brust, G.E. & House, G.J. (1988) Weed seed destruction by arthropods and rodents in low-input soybean agroecosystems. American Journal of Alternative Agriculture, 3, 1925.
  • Byers, R.A. & Barratt, B.I.P. (1991) Behavior of slugs, Derocerus reticulatum (Gastropoda, Limicidae) and crickets, Gryllus pennslyvanicus (Orthoptera, Gryllidae), on seedling alfalfa. Entomological News, 102, 200204.
  • Cardina, J., Norquay, H.M., Stinner, B.R. & McCartney, D.A. (1996) Postdispersal predation of velvetleaf (Abutilon theophrasti) seeds. Weed Science, 44, 534539.
  • Carmona, D.M., Menalled, F.D. & Landis, D.A. (1999) Gryllus pennslyvanicus (Orthoptera: Gryllidae): laboratory weed seed predation and within field activity–density. Journal of Economic Entomology, 92, 825829.
  • Clark, W.R. & Young R.E. (1986) Crop damage by small mammals in no-till cornfields. Journal of Soil and Water Conservation, 41, 338341.
  • Cousens, R.D., Dale, M.R.T., Taylor, J., Law, R., Moerkerk, M. & Kembel, S.W. (2006) Causes of pattern in plant communities where environmental change is rapid and species longevity is short. Journal of Vegetation Science, 17, 509524.
  • Cromar, H.E., Murphy, S.D. & Swanton, C.J. (1999) Influence of tillage and crop residue on postdispersal predation of weed seeds. Weed Science, 47, 184194.
  • Davis, A.S., Dixon, P.M. & Liebman M. (2003) Cropping system effects on giant foxtail demography: II. Retrospective perturbation analysis. Weed Science, 51, 930939.
  • Douglass, R.J. (1989) The use of radio-telemetry to evaluate microhabitat selection by deer mice. Journal of Mammalogy, 70, 648652.
  • Forcella, F., Peterson, D.H. & Barbour, J.C. (1996) Timing and measurement of weed seed shed in corn (Zea mays). Weed Technology, 10, 535543.
  • Forrester, G.E. & Steele, M.A. (2000) Variation in the presence and cause of density-dependent mortality in three species of reef fishes. Ecology, 81, 24162427.
  • Genstat 5 Committee (1993) genstat Release 3 Reference Manual. Clarendon Press, Oxford, UK.
  • Getz, L.L. & Brighty, E. (1986) Potential effects of small mammals in high-intensity agricultural systems in east-central Illinois, USA. Agriculture, Ecosystems and Environment, 15, 3950.
  • González-Andujar, J.L. & Fernandez-Quintanilla, C. (1991) Modelling the population dynamics of Avena sterilis under dry-land cereal cropping systems. Journal of Applied Ecology, 28, 1627.
  • Harrison, S.K., Regnier, E.E. & Schmoll, J.T. (2003) Postdispersal predation of giant ragweed (Ambrosia trifida) seed in no-tillage corn. Weed Science, 51, 955964.
  • Heads, P.A. & Lawton, J.H. (1983) Studies on the natural enemy complex of the holy leaf-miner: the effects of scale on the detection of aggregative responses and the implications for biological control. Oikos, 105, 657663.
  • Heggenstaller, A.H., Menalled, F.D., Liebman, M. & Westerman, P.R. (2006) Seasonal patterns in post-dispersal seed predation of Abutilon theophrasti and Setaria faberi in three cropping systems. Journal of Applied Ecology, 43, 9991010.
  • Holling, C.S. (1959) The components of predation as revealed by a study of small-mammal predation of the European pine sawfly. Canadian Entomologist, 91, 293320.
  • Honřk, A., Martinková, Z. & Saska, P. (2005) Post-dispersal predation of Taraxacum officinale (Dandelion) seed. Journal of Ecology, 93, 345352.
  • Hough-Goldstein, J.A., VanGessel, M.J. & Wilson, A.P. (2004) Manipulation of weed communities to enhance ground-dwelling arthropod populations in herbicide-resistant field corn. Environmental Entomology, 33, 577586.
  • Hulme, P.E. (1993) Post-dispersal seed predation by small mammals. Symposia of the Zoological Society of London, 65, 269287.
  • Hulme, P.E. (1994) Post-dispersal seed predation in grassland: its magnitude and sources of variation. Journal of Ecology, 82, 645652.
  • Hulme, P.E. (1997) Post-dispersal seed predation and the establishment of vertebrate dispersed plants in Mediterranean scrublands. Oecologia, 111, 9198.
  • Hulme, P.E. & Hunt, M.K. (1999) Rodent post-dispersal seed predation in deciduous woodland: predator response to absolute and relative abundance of prey. Journal of Animal Ecology, 68, 417428.
  • Jordan, N., Mortensen, D.A., Prenzlow, D.M. & Cox, K.C. (1995) Simulation analysis of crop rotation effects on weed seedbanks. American Journal of Botany, 82, 390398.
  • Kirk, V.M. (1972) Seed-caching by larvae of two ground beetles, Harpalus pensylvanicus and H. erraticus. Annals of the Entomological Society of America, 65, 14261428.
  • Kollmann, J. & Bassin, S. (2001) Effects of management on seed predation in wildflower strips in northern Switzerland. Agriculture, Ecosystems and Environment, 83, 285296.
  • Leguizamón, E.S. & Roberts, H.A. (1982) Seed production by an arable community. Weed Research, 22, 3539.
  • Lund, R.D. & Turpin, F.T. (1977) Carabid damage to weed seeds found in Indiana cornfields. Environmental Entomology, 6, 695698.
  • Manson, R.H. & Stiles, E.W. (1998) Links between microhabitat preferences and seed predation by small mammals in old fields. Oikos, 82, 3750.
  • Marino, P.C., Westerman, P.R., Pinkert, C. & Werf, W van der (2005). Influence of seed density and aggregation on post-dispersal weed seed predation in cereal fields. Agriculture, Ecosystems and Environments, 106, 1725.
  • McShea, W.J. & Gilles, A.B. (1992) A comparison of traps and fluorescent powder to describe foraging for mast by Peromyscus leucopus. Journal of Mammalogy, 73, 218222.
  • Moorcroft, D., Whittingham, M.J., Bradbury, R.B. & Wilson, J.D. (2002) The selection of stubble fields by wintering granivorous birds reflects vegetation cover and food abundance. Journal of Applied Ecology, 39, 535547.
  • Neubert, M.G. & Caswell, H. (2000) Density-dependent rates and their population dynamic consequences. Journal of Mathematical Biology, 41, 103121.
  • O’Rourke, M.E., Heggenstaller, A.H., Liebman, M. & Rice, M.E. (2006) Post-dispersal weed seed predation by invertebrates in conventional and low-external-input crop rotation systems. Agriculture, Ecosystems and Environments, 116, 280288.
  • Parmenter, R.R. & MacMahon, J.A. (1988) Factors influencing species composition and population sizes in a ground beetle community (Carabidae): predation by rodents. Oikos, 52, 350356.
  • Plesner Jensen, S. & Honess, P. (1995) The influence of moonlight on vegetation height preference and trappability of small mammals. Mammalia, 59, 3542.
  • Price, M.V. (1977) Validity of live trapping as a measure of foraging activity of heteromyid rodents. Journal of Mammalogy, 58, 107110.
  • Reed, A.W., Kaufman, G.A. & Kaufman, D.W. (2004) Influence of fire, topography, and consumer abundance on seed predation in tallgrass prairie. Canadian Journal of Zoology, 82, 14591467.
  • Ruscoe, W.A., Elkington, J.S., Choquenot, D. & Allen, R.B. (2005) Predation of beech seed by mice: effects of numerical and functional responses. Journal of Animal Ecology, 74, 10051019.
  • Sunderland, K.D., De Snoo, G.R., Dinter, A., Hance, T., Helenius, J., Jepson, P., Kromp, B., Lys, J.A., Samu, F., Sotherton, N.W., Toft, S. & Ulber, B. (1995) Density estimation for in vertebrate predators in agroecosystems. Arthropod natural enemies in arable land. I. Density, Spatial Heterogeneity and Dispersal (eds S.Toft & W.Riedel), pp. 133162. Aarhus University Press, Aarhus, Denmark.
  • Tew, T.E., Todd, I.A. & Macdonald, D.W. (2000) Arable habitat use by wood mice (Apodemus sylvaticus). 2. Microhabitat. Journal of the Zoology (London), 250, 305311.
  • Turchin, P. (1995) Population regulation: old arguments and a new synthesis. Population Dynamics, New Approaches and Synthesis (eds N.Cappuccina & P.W.Price), pp. 1940. Academic Press, San Diego, California.
  • Westerman, P.R., Liebman, M., Heggenstaller, A.H. & Forcella, F. (2006) Integrating measurements of seed availability and removal to estimate weed seed losses due to predation. Weed Science, 54, 566574.
  • Westerman, P.R., Wes, J.S., Kropff, M.J. & Werf, W. van der (2003) Annual losses of weed seeds due to predation in organic cereal fields. Journal of Applied Ecology, 40, 824836.
  • Wiles, L.J., Barlin, D.H., Schweitzer, E.E., Duke H.R. & Whitt, D.E. (1996) A new soil sampler and elutriator for collecting and extracting weed seeds from soil. Weed Technology, 10, 3541.
  • Zhang, J., Drummond, F.A., Liebman, M. & Hartke, A. (1997a) Phenology and dispersal of Harpalus rufipes DeGeer (Coleoptera: Carabidae) in agroecosystems in Maine. Journal of Agricultural Entomology, 14, 171186.
  • Zhang, J., Drummond, F.A., Liebman, M. & Hartke, A. (1997b) Insect predation of seeds and plant population dynamics. Maine Agricultural and Forest Experimental Station, Technical Bulletin, 163, 28.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Appendix S1. Crop management.

Appendix S2. Calculations of initial seed densities.

Table S1. Summary of corrections made to the initial seed density.

Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JPE_1481_sm_AppendixS1.doc35KSupporting info item
JPE_1481_sm_AppendixS2.doc35KSupporting info item
JPE_1481_sm_TableS1.doc55KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.