Introduction
The need for management of weed seedbanks is widely recognized (Davis et al. Reference Davis, Cardina, Forcella, Johnson, Kegode, Lindquist, Luschei, Renner, Sprague and Williams2005; Gallandt et al. Reference Gallandt, Liebman and Huggins1999; Jordan et al. Reference Jordan, Mortensen, Prenzlow and Cox1995; Liebman and Mohler 2001). Many studies have shown a decrease in weed seed density in the soil in response to tillage (Lueschen et al. Reference Lueschen, Andersen, Hoverstad and Kanne1993; Roberts and Feast Reference Roberts and Feast1972; Schweizer and Zimdahl Reference Schweizer and Zimdahl1984), and to date, tillage coupled with the elimination of seed rain has been the most effective means for decreasing weed seedbanks. However, because of potential soil degradation associated with frequent tillage, other methods are needed for reducing weed seedbanks.
Manipulation of bacterial, fungal, and macrofaunal populations that attack seeds in the soil offers a potential alternative to tillage as a means for managing weed seedbanks (Gallandt et al. Reference Gallandt, Molloy, Lynch and Drummond2005; Kremer Reference Kremer1993). Soil has been experimentally inoculated with bacterial and fungal populations known to attack weed seeds (Kremer Reference Kremer1993), but this approach faces the many problems associated with bioherbicides (Auld et al. Reference Auld, Hetherington and Smith2003; Ghosheh Reference Ghosheh2005). A more promising approach is to use cover crops, later incorporated as green manures, to promote animal and microbial populations that attack weed seeds. While the cover crop is growing, it can provide habitat beneficial to seed predators (Gallandt et al. Reference Gallandt, Molloy, Lynch and Drummond2005). Once incorporated, the cover crop residue feeds microorganisms that could potentially attack weed seeds. However, studies comparing cropping systems with and without green manures have shown contradictory effects on weed seed populations. Fennimore and Jackson (Reference Fennimore and Jackson2003) found reduced burning nettle (Urtica urens L.) and shepherd’s purse [Capsella bursa-pastoris (L.) Medik.] seedling densities in treatments amended with green manures and compost compared with control treatments. They further found lower densities of C. bursa-pastoris seeds in the amended treatments and a negative correlation between microbial biomass and density of U. urens and C. bursa-pastoris seedlings. Similarly, Dyck and Liebman (Reference Dyck and Liebman1994) found lower emergence of common lambsquarters (Chenopodium album L.) following incorporation of crimson clover (Trifolium incarnatum L.). The reduced emergence could potentially have been due to either suppression of germination or to increased death of seeds and of seedlings before emergence. Suppression of germination is unlikely, however, because nitrate released during decomposition of a legume would tend to promote germination of C. album (Henson Reference Henson1970; Roberts and Benjamin Reference Roberts and Benjamin1979; Vincent and Roberts Reference Vincent and Roberts1977). Mohler et al. (Reference Mohler, Dykeman, Nelson and DiTommaso2014) found that incorporation of pea (Pisum sativum L.) reduced emergence of several common weed species and that this was associated with infection of seeds by Fusarium species.
In contrast with these studies, Davis et al. (Reference Davis, Anderson, Hallett and Renner2006) found that after 10 yr of cropping system history, seed mortality of velvetleaf (Abutilon theophrasti Medik.) and giant foxtail (Setaria faberi Herrm.) was greater in conventional systems than in a system that received only compost as a nutrient source. Moreover, S. faberi mortality was also greater in a conventional system than in a system with red clover (Trifolium pratense L.) green manure. They also showed that seed mortality rates were correlated with the composition of the microbial community. Davis (Reference Davis2007) found that addition of corn stover to soil had no effect on seed mortality of five species and decreased the mortality of A. theophrasti compared with a control without corn stover. For giant ragweed (Ambrosia trifida L.) and wooly cupgrass [Eriochloa villosa (Thunb.) Kunth], corn stover increased or decreased weed persistence, depending on whether or not inorganic nitrogen was added. Thus, the overall addition of organic matter mostly either had no effect on seed mortality or decreased mortality. Similarly, Gallandt et al. (Reference Gallandt, Fuerst and Kennedy2004) found that whether crop residues were left on the soil surface in a no-tillage system or incorporated in a conservation tillage system had no effect on decay of wild oat (Avena fatua L.) seeds.
Incorporated cover crops can potentially affect weed seed populations in three ways. First, cover crops can attract macrofauna like earthworms that consume weed seeds. Earthworm population density is correlated with soil carbon (Edwards Reference Edwards1983), and earthworms consume weed seeds (McRill and Sagar Reference McRill and Sagar1973; Shumway and Koide Reference Shumway and Koide1994). Second, incorporation of cover crop residue promotes growth of microbial populations, which could potentially damage weed seeds. This was the focus of much of the work discussed earlier.
Finally, incorporated cover crops can potentially promote or inhibit germination of weed seeds, thereby decreasing or increasing the persistence of seedbanks. If the seed germinates but the seedling dies before emergence because it is too deep in the soil or is killed by some agency, then this fatal germination is essentially indistinguishable from seed mortality. Legume cover crop residue generally has a low carbon to nitrogen ratio, and consequently, decomposition of this material releases nitrogen, which is quickly converted to nitrate. About half of all weed species have increased germination under at least some conditions when nitrate is present in the germination environment (Steinbauer and Grigsby Reference Steinbauer and Grigsby1957). Thus, an incorporated legume could promote germination and thereby reduce seed persistence of nitrate-sensitive species. Conversely, incorporation of materials with a high carbon to nitrogen ratio, such as nearly mature cereal grain and grain straw, can be expected to promote microbial sequestration of nitrogen, thereby reducing potential germination in response to nitrate released during decomposition of soil organic matter. Similarly, allelopathic chemicals released from decomposing crop residue sometimes inhibit or promote germination (Altieri and Doll Reference Altieri and Doll1978; Chung and Miller Reference Chung and Miller1995; Reigosa et al. Reference Reigosa, Souto and Gonzalez1999), and that could affect seed persistence. Allelopathic chemicals rarely kill seeds before germination but can act as one of the several factors preventing successful emergence. Many factors associated with tillage promote germination, including exposure to light (Sauer and Struik Reference Sauer and Struik1964), increased diurnal temperature fluctuations (Fausey and Renner Reference Fausey and Renner1997; Henson Reference Henson1970), and the venting of volatile germination inhibitors (Holm Reference Holm1972), and these processes can deplete seedbanks. However, these processes will act about equally whether residues are incorporated during tillage or not. The present study explored the second and third possible source of changes in seed persistence associated with the incorporation of cover crops. Because the experiment was conducted with the seeds confined in bags, macrofauna were excluded.
Materials and Methods
The experiment was conducted in a Cornell University Agricultural Experiment Station field adjacent to the Cornell University campus in Ithaca, NY (42.450°N, 76.460°W). The soil was a Williamson very fine sandy loam (coarse-silty, mixed, active, mesic, Typic Fragiudepts). The field had been fallow before the experiment. On August 23, 2011, the field was sprayed with glufosinate ammonium at 0.25 kg ai ha−1, mowed 8 d later, disked, and on September 1, 2011, planted with cover crops. The experiment was laid out in a replicated block design with five replications and three treatments: bare control, rye (Secale cereale L.), and hairy vetch (Vicia villosa Roth.). ‘Aroostook’ rye was drilled at 123 kg ha−1 and hairy vetch at 67 kg ha−1 in this and subsequent years using a Great Plains model 3P1006NT grain drill (Great Plains Manufacturing, Salina, KS). Plots were 3.0 by 13.7 m.
The basic approach of the experiment was to bury bags of weed seeds mixed with sand in the late fall when cool soil temperatures prevented germination. Each succeeding spring for the next 3 yr, all bags were exhumed and opened. Some samples were retained for testing; the remaining samples had chopped cover crop material mixed with the sand and seeds or, in the case of the control treatment, the sample was simply stirred. The samples were then rebagged and buried until the next spring. This procedure was intended to simulate spring tillage for incorporation of cover crops. The entire experiment was conducted twice, with initial burials in 2011 and 2013.
Seeds of C. album, Powell amaranth (Amaranthus powellii S. Watson), and S. faberi were collected from agricultural populations in central New York state. Mature whole plants of C. album and A. powellii were cut near the plant base and allowed to dry at room temperature in large paper bags. Ripe seeds of S. faberi were shaken from the inflorescences into a bucket and allowed to further dry at room temperature. Fully black capsules of A. theophrasti were collected from buffer areas of a nutrient rate experiment (Little et al. 2014) and allowed to dry at room temperature. When seeds were judged to be dry, they were threshed out from other plant material and thoroughly cleaned. Seeds were collected the same fall they were buried, except that the 2011 A. theophrasti seeds were also used in 2013. These were stored dry at 4 C until used.
Seeds were counted into coin envelopes using a Model 801-7/B Seedburo seed counter (Seedburo Equipment, Des Plaines, IL) with an accuracy of 1% to 3% (Table 1). Just before burial, seeds were mixed with 57 ml (83 g) of silica sand that had passed a 0.5-mm sieve and were placed into 7.5 by 10 cm organza bags (Uline, Pleasant Prairie, WI). The bags were stapled shut and placed into a second bag, and this too was stapled shut. Bags were tied to approximately 40 cm of nylon cord. A plastic label stating the species, treatment, replication, and year to be recovered for testing was threaded onto the cord, and the end was tied to three 2.5-cm (inside diameter) steel washers. Seed bags were buried initially on October 31, 2011, and November 21, 2013, at a depth of 15 cm in individual holes. Holes were made with an 18 by 7 cm diameter bulb planter. The washers were placed at the bottom of the hole, then the plastic tag, a dusting of soil, and finally the bag containing the sand and seeds. Holes were spaced 90 cm apart along a line 1 m from the plot edge to avoid wheel traffic.
Table 1 Number of seeds per bag.
In mid-May of each year from 2012 through 2015, weeds in the control treatments were sprayed with glyphosate (1.5 kg ai ha−1) or hoed, and hairy vetch or rye plants that were in the wrong plot due to imprecision during planting were removed to bare spots in plots of the appropriate treatment. A few days before the bags were to be excavated, aboveground cover crop biomass was sampled by clipping and separately bagging cover crops and weeds in two randomly placed 33 by 76 cm (0.25-m−2) quadrats. This material was weighed, dried for 3 d at 60 C, and weighed again. This provided both an estimate of cover crop biomass and the ratio of fresh to dry weight. Additional cover crop material was collected and chopped into approximately 0.5-cm pieces. The chopped material was weighed out into small plastic bags that contained the amount of material corresponding to the incorporation of 600 dry g m−2 of cover crop incorporated to a depth of 15 cm that would be contained in 57 ml of soil (the volume of sand in the bag). Previous experiments have shown that 600 gm−2 of aboveground cover crop material is about the maximum that can be grown between September and late May in central New York state.
During the last week in May of each year, cover crops were mowed with a Mott 74 flail mower (Alamo Industrial, Seguin, TX). Each bag was located with a metal detector and marked with a flag. The next day, all bags were exhumed. Bags were immediately placed in an ice chest to keep them at close to soil temperature. In the laboratory, soil was brushed off of the bags, and bags were opened into a pan. Samples that were destined for reburial were stirred (control) or mixed with the appropriate type of cover crop material, rebagged, and returned to an ice chest. While samples were being processed, the field plots were tilled to a depth of about 12 cm with a Maschio 155 rotary tiller (Machio Gaspardo North America, DeWitt, IA) to incorporate cover crops. The next day, bags were reburied as before. Samples that were to be analyzed for seed viability were spread in 15-cm aluminum pie pans and dried for 24 h at 40 C. The seeds were then separated from sand by sieving and stored at 5 C and 30% relative humidity until testing.
Sorghum-sudangrass [Sorghum bicolor (L.) Moench. ssp. drummondi (Nees ex Steud.) de Wet & Harlan] was planted at a rate of 73 kg ha−1 on the whole field just before or after the bags were reburied. Sorghum-sudangrass was harvested using a John Deere 972 forage chopper (John Deere, Moline, IL) in early September and removed from the field. The plots were sprayed with glyphosate at 2.3 kg ai ha−1, and the cover crop plots were no-till planted with rye and hairy vetch, as in 2011.
The original seed lots of each species were tested at selected temperature and light regimens to determine which gave the highest percentage of germination. All species were tested on two layers of moistened blue germination blotters (Anchor Paper, St Paul, MN) in 9 by 9 by 3.5 cm germination boxes. Germination tests were performed in a growth chamber (Model I-36LL, Percival Scientific, Perry, IA). Germination of A. theophrasti was tested at 35 C in the dark for 7 d. Dormancy in A. theophrasti is maintained by an impermeable seed coat (Horowitz and Taylorson Reference Horowitz and Taylorson1985), so viability of hard seeds was then determined by piercing the hilum with a needle and allowing the seeds to germinate for another 7 d.
Germination of A. powellii seeds was tested with 14/10 h light/dark at 35/20 C for 14 d (Weaver and Thomas Reference Weaver and Thomas1986). Germination of C. album and S. faberi was tested with 8/16 h light/dark at 30/20 C for 14 d. Viability of nongerminated C. album and A. powellii seeds was evaluated by staining with a 1.0 % solution of 2,3,5-triphenyl tetrazolium chloride (USB, Cleveland, OH), and S. faberi viability was evaluated using a 0.1% tetrazolium solution(AOSA/SCST 2010). Viability of the original seed lots was tested using a minimum of five replicates of 50 seeds each. Tests on recovered seeds were done on 50 randomly selected seeds from each replicate bag unless fewer than 50 seeds were present, in which case all seeds were tested.
Data Analysis
The numbers of viable seeds (germinable plus dormant) present in succeeding years were analyzed with general linear mixed models assuming a Poisson distribution with overdispersion (SAS PROC GLIMMIX, Statistical Analysis System, Cary, NC). Treatment, year, and their interaction were treated as fixed effects, and replication and treatment within replication were treated as random effects. To allow for overdispersion, the interaction between treatment, replication, and year was also treated as a random effect. The significance of difference between treatments of the change in number of viable seeds in succeeding years was determined by predefined contrasts and was used to indicate the significance of differences in seed persistence between treatments. Since the algorithm used a natural log link, estimates of the difference between years in least-squares means were back-transformed by exponentiation to provide estimates of seed persistence.
Results and Discussion
Incorporation of hairy vetch produced substantial reductions in the survival of C. album and A. powellii over the 2.5-yr period from initial burial to final exhumation (Table 2). For these species, seed persistence was reduced two-thirds to 1/4 that of the control treatment. The decreased persistence of C. album and A. powellii seeds in the presence of decomposing hairy vetch relative to the control could be related to the sensitivity of seeds of these two species to nitrate in the soil solution. Hairy vetch is a nitrogen-fixing legume and tissues have a high nitrogen content and release substantial amounts of nitrogen during decomposition (Sarrantonio and Scott Reference Sarrantonio and Scott1988). As the vetch decomposes, it releases this nitrogen, which is quickly oxidized to nitrate by soil bacteria. Weak solutions of nitrate have been shown to increase the germination of C. album (Henson Reference Henson1970; Roberts and Benjamin Reference Roberts and Benjamin1979; Vincent and Roberts Reference Vincent and Roberts1977) and A. powellii and related pigweed (Amaranthus) species (Brainard et al. Reference Brainard, DiTommaso and Mohler2006; Costea et al. Reference Costea, Weaver and Tardif2004). In the present experiment, seeds were placed too deeply in the soil for successful emergence of any of the four species, even if they had not been confined to bags. Thus, in the experiment, any nitrate-promoted germination was fatal and a similar promotion of fatal germination could be expected to work against seeds of C. album and A. powellii exposed to decomposing hairy vetch in the soil of a farmer’s field. One might argue that the nitrate released from decomposing hairy vetch might also promote germination that leads to greater emergence of these species. However, we have demonstrated elsewhere that incorporated legumes decrease emergence of these and other species, probably by promoting a short-term increase in pathogenic fungi (Mohler et al. Reference Mohler, Dykeman, Nelson and DiTommaso2014). In results similar to ours, Hill et al. (Reference Hill, Renner, Sprague and Davis2016) found that seeds of C. album buried with red clover residue were less persistent than seeds buried without residue in 1 of 2 yr. They showed that the clover residue released substantial amounts of inorganic nitrogen. This nitrogen could have triggered fatal germination of C. album.
Table 2 Persistence rates of weed seeds buried at two times, fall 2011 and fall 2013, and recovered for analysis 2.5 yr later following additions of rye, hairy vetch, or no (control) cover crop residue each subsequent spring.Footnote a
a Abbreviations: C, control; F, fall; R, rye; S, spring; V, hairy vetch.
b P values for predefined comparisons that are significant (P<0.05) or nearly significant (0.05<P<0.1) are shown in bold.
In contrast with C. album and A. powellii, A. theophrasti and S. faberi showed no significant decrease in persistence associated with incorporation of hairy vetch relative to the control. Germination of A. theophrasti is not promoted by nitrate (Fawcett and Slife Reference Fawcett and Slife1978; Sweeney et al. Reference Sweeney, Renner, Laboski and Davis2008). Consequently, fatal germination in response to nitrate released from decaying hairy vetch would not be expected to affect persistence of A. theophrasti. Setaria faberi similarly showed no significant reduction in persistence with hairy vetch relative to the control. However, apparent persistence of S. faberi with hairy vetch was only 36% that for the control, so the lack of a significant response to hairy vetch may be due to insufficient sample size in the face of extreme variability. Data on the germination response of S. faberi to nitrate are ambiguous. Fawcett and Slife (Reference Fawcett and Slife1978) found no response of S. faberi emergence to applications of nitrogen. Sweeney et al. (Reference Sweeney, Renner, Laboski and Davis2008) did observe increased emergence of sown S. faberi following nitrogen application but found no increase in emergence of S. faberi from the seedbank in response to nitrogen. Schimpf and Palmblad (Reference Schimpf and Palmblad1980) found a weak germination response to nitrogen for fresh seeds, but no response to nitrogen for seeds that had been chilled. Dekker (Reference Dekker2003) reviewed the literature on germination responses of foxtails (Setaria spp.) but did not mention any effect of nitrogen on germination. He did note, however, that germination responses of all species in the genus are highly variable and change with environmental conditions. In a study similar to ours, Hill et al. (Reference Hill, Renner, Sprague and Davis2016) found that incorporated red clover had no effect on persistence of S. faberi. We conclude that incorporated legume cover crops are likely to have a weak or inconsistent effect on the persistence of S. faberi.
Because samples were collected for analysis at multiple times following the initial seed burial, additional insights can be gained by examining seed persistence over specific intervals (Table 3). The number of viable seeds varied greatly among replicates, indicating that seed survival rates are strongly affected by microvariation in soil conditions. Occasionally, this variation led to a few more seeds being present in a later sample than were present in the sample drawn from that plot the previous year. The general linear model algorithm, however, handled such cases without problems. For two species by interval cases, the later year had consistently more viable seeds than the previous year. This situation could not occur if the number of viable seeds present in a later sampling year were determined primarily by seed survival. Specifically, more A. theophrasti seeds were present for all plots in 2013 than 2012, and more S. faberi seeds were present in most samples in 2015 than 2014. As such situations could only occur due to some problem with handling or storage of the seeds during or after exhumation, the data for the earlier of the 2 yr were discarded for these cases.
Table 3 Persistence rates of weed seeds buried at two times, fall 2011 and fall 2013, and recovered for analysis the following three springs.Footnote a
a Abbreviations: C, control; F, fall; R, rye; S, spring; V, hairy vetch; NA, data not available.
b P values for predefined comparisons are given below the proportion of seeds persisting during the interval. P-values that are significant (P < 0.05) or nearly significant (0.05 < P < 0.1) are shown in bold.
c Survival rates of 1.00 were substituted for estimates slightly greater than 1 for A. theophrasti in the control treatment for S2014–S2015 and A. theophrasti in the hairy vetch treatment for S2015–S2016.
Treatment had no effect on seed persistence during the first winter the seeds were in the soil (Table 3). This was as expected, because no cover crop material had been incorporated into the bags at the initial burial. Some fine roots did grow into the bags, and these could have potentially affected seed persistence. However, the degree of root penetration into the bags varied greatly between bags, apparently without pattern. Consequently, any effect the cover crop roots may have had on seed persistence was masked by the high variability in persistence.
Large reductions in seed persistence relative to the controls occurred for C. album following the first incorporation of hairy vetch, that is, from 2012 to 2013 for the 2011 initial burial and 2014 to 2015 for the 2013 initial burial (Table 3). Hairy vetch had no effect on C. album persistence following the second input of hairy vetch residue. Only a portion of the seeds in any seed lot of C. album are sensitive to nitrate (Henson Reference Henson1970; Roberts and Benjamin Reference Roberts and Benjamin1979; Vincent and Roberts Reference Vincent and Roberts1977). Possibly, the nitrate-sensitive individuals germinated following the first incorporation of hairy vetch and thus were not present to be affected by the second incorporation.
In contrast with C. album, the reduction in A. powellii seed persistence with hairy vetch compared with the control followed the second incorporation of hairy vetch rather than the first. Although the reduction in seed persistence in the 2015 to 2016 interval did not quite reach statistical significance (P = 0.08), only 21% as many seeds persisted in the presence of hairy vetch compared with the control. Conceivably, the A. powellii seeds aged into a sensitivity to nitrate, but the literature provides no basis for this hypothesis. Consequently, why the effect of the hairy vetch was delayed until the second addition of residue is unknown.
The effect of incorporated rye residue on seed persistence was mostly negligible and without obvious pattern when it did occur. Rye about doubled persistence relative to the control of S. faberi from fall 2011 to spring 2014 and C. album from fall 2013 to spring 2016. This increase in persistence due to rye is similar, though less consistent, to that observed by Davis (Reference Davis2007) when corn stover, another residue with high carbon to nitrogen ratio, was incorporated. Balancing the observed increase in persistence due to rye for S. faberi and C. album, however, was a decrease in persistence in the presence of rye relative to the control for A. theophrasti from 2014 to 2015 and A. powellii from 2012 to 2013. Unlike the effects of hairy vetch on persistence of C. album and A. powellii, the effects of rye over the 2.5-yr intervals (Table 2) are not easily related to persistence in the three shorter intervals (Table 3).
The frequently negligible and sometimes positive effect of rye on seed persistence and the lack of an effect of hairy vetch on the two species with low or no sensitivity to nitrate indicates that attempting to promote microbial attack on weed seeds by incorporation of cover crop residue may be a poor strategy for reducing weed seedbanks. However, we do not discount the potential importance of microbes in affecting the persistence of seeds in the soil. The density and composition of the microbial community probably varies greatly on small scales within the soil and may contribute to the high variability in seed persistence between replicates observed in this study. Nevertheless, the present study, in conjunction with published literature, indicates that a general promotion of microbial activity through the incorporation of cover crops is not an effective method for the management of weed seedbanks.
Because we did not analyze microbial communities, we cannot rule out the possibility that microbial attack rather than fatal germination was responsible for the reduced persistence of C. album and A. powellii following incorporation of hairy vetch. We believe, however, that fatal germination is the more parsimonious explanation of the effect. Whether depletion of seedbanks of nitrate-sensitive species through incorporation of legume cover crops can be broadly applied in the field remains to be determined. Previous studies examined seedling emergence of nitrate-sensitive species following application of nitrate fertilizers and found no response or variable responses among years, weed populations, seed lots, and even position of seeds on the parent plant (Brainard et al. Reference Brainard, DiTommaso and Mohler2006; Fawcett and Slife Reference Fawcett and Slife1978; Hilton Reference Hilton1985; Hurtt and Taylorson Reference Hurtt and Taylorson1986). Although these studies focused on emergence, the effect or lack of effect of nitrate was presumably on germination rather than on seedling growth through the soil, and thus these studies are relevant to persistence of seedbanks. Nitrate sensitivity can also change as seeds age, are chilled over the winter (Schimpf and Palmblad Reference Schimpf and Palmblad1980), or are affected by other environmental factors (Hilton Reference Hilton1984). Clearly, additional studies are needed to explore the causes and consistency of the decreased persistence of some species of weed seeds following incorporation of legume cover crops. Nevertheless, promotion of fatal germination of seeds of nitrate-sensitive species through incorporation of legumes may reduce weed seedbanks in some circumstances.
Acknowledgements
We thank S. Morris, P. Putriment, E. Kennedy, K. Ellis, X. Garcia Friere. K. Bitter, M. Harbert, V. Minares, D. Stauffer, M. Loos, S. Costello, C. Catranis, W. Huang. K. Walsh, J. Hunn, Q. Maqsood, Z. Kaufmann, M. Amirkhani, N. Brennan, A. Repka, C. Stokes, L. Woodruff, C. Jennison, E. Marino, M. Dias, P. Kikhao, J. Drennan, J. France, P. Tremel, D. Stauffer, F. Kordbacheh, C. Neumann, and E. Buck for assistance with the lab work and fieldwork, and R. J. Richtmyer for conducting the machinery operations. This work was supported by funds from the Cornell University and New York State Agricultural Experiment Station (Multi-State Hatch Projects NE-1026, and NE-1047, Project numbers 3120006036 NYC-145843 NE1026, and 231327 NYC-125800 NE1047). No conflicts of interest have been declared.
