Seed dispersal has key implications for community dynamics and restoration ecology. However, estimating seed rain (the number and diversity of seeds arriving in a given area) is challenging, and the lack of standardization in measurement prevents cross-site comparisons. Seed trap effectiveness and accuracy of seed sorting methods are key components of seed rain estimates in need of standardization. We propose and describe a standardized protocol for evaluating the effectiveness of two seed trap types (sticky and funnel traps) and the accuracy of a seed sorting method. We used widely available seeds (arugula, quinoa, sesame and sunflower) to produce a gradient of seed size, weight and colour. Proof-of-concept was tested in a tropical grassland, where traps were set for 30 days. Our results suggest that we underestimate dispersal of seeds with less than 2 mm width that can be easily mistaken for debris and soil particles or that fail to adhere to sticky traps. Seeds on sticky traps may be more vulnerable to removal by wind and rain, whereas seeds in funnel traps are more susceptible to decay. We found no evidence of observer bias on seed sorting for funnel trap samples. However, accuracy on seed sorting for funnel trap samples tended to decline for seeds with less than 2 mm width, suggesting a size-dependence in seed retrieval success. Our standardized protocol addressing trap effectiveness and seed sorting methods will increase the reliability of data obtained in seed rain studies and allow more reliable comparisons between datasets.

Potential effectiveness of harvest weed seed control (HWSC) systems depends upon seed shatter of the target weed species at crop maturity, enabling its collection and processing at crop harvest. However, seed retention likely is influenced by agroecological and environmental factors. In 2016 and 2017, we assessed seed-shatter phenology in 13 economically important broadleaf weed species in soybean [Glycine max (L.) Merr.] from crop physiological maturity to 4 wk after physiological maturity at multiple sites spread across 14 states in the southern, northern, and mid-Atlantic United States. Greater proportions of seeds were retained by weeds in southern latitudes and shatter rate increased at northern latitudes. Amaranthus spp. seed shatter was low (0% to 2%), whereas shatter varied widely in common ragweed (Ambrosia artemisiifolia L.) (2% to 90%) over the weeks following soybean physiological maturity. Overall, the broadleaf species studied shattered less than 10% of their seeds by soybean harvest. Our results suggest that some of the broadleaf species with greater seed retention rates in the weeks following soybean physiological maturity may be good candidates for HWSC.

Seed shatter is an important weediness trait on which the efficacy of harvest weed seed control (HWSC) depends. The level of seed shatter in a species is likely influenced by agroecological and environmental factors. In 2016 and 2017, we assessed seed shatter of eight economically important grass weed species in soybean [Glycine max (L.) Merr.] from crop physiological maturity to 4 wk after maturity at multiple sites spread across 11 states in the southern, northern, and mid-Atlantic United States. From soybean maturity to 4 wk after maturity, cumulative percent seed shatter was lowest in the southern U.S. regions and increased moving north through the states. At soybean maturity, the percent of seed shatter ranged from 1% to 70%. That range had shifted to 5% to 100% (mean: 42%) by 25 d after soybean maturity. There were considerable differences in seed-shatter onset and rate of progression between sites and years in some species that could impact their susceptibility to HWSC. Our results suggest that many summer annual grass species are likely not ideal candidates for HWSC, although HWSC could substantially reduce their seed output during certain years.

The development of early-maturing soybean [Glycine max (L.) Merr.] varieties has led to an increase in soybean production in canola (Brassica napus L.)-dominant crop rotations in western Canada. Herbicide-resistant (HR) volunteer B. napus can be difficult to manage in HR soybean using herbicides alone. In 2013 and 2014, four field experiments were conducted in Manitoba, Canada, to evaluate soybean row spacing, seeding density, nitrogen supply, and interrow tillage as candidate nonchemical weed management tools for an integrated program to manage volunteer B. napus in soybean. Among treatments and sites, volunteer B. napus produced about 830 seeds plant−1 and resulted in large seedbank inputs (averaging about 20,300 seeds m−2). Volunteer B. napus seedling recruitment differed among sites, and resulted in two distinct classes of sites based on average seedling densities of 39 and 89 plants m−2. Weed management tools were more effective at the sites with higher volunteer B. napus densities. At these sites, soybean yield was greater when using an increased soybean-seeding density (44% greater yield using a seeding density of 682,500 vs. 455,000 seeds ha−1) or interrow tillage (36% greater yield with vs. without using interrow tillage). Soybean row spacing (19 vs. 38 vs. 76 cm) did not affect soybean yield, unless the reduction in row spacing was combined with an increased seeding density (65% greater yield with narrow-row soybean seeded at 682,500 vs. wide-row soybean seeded at 455,000 seeds ha−1). At the sites with higher volunteer B. napus densities, seed production of canola volunteer B. napus was greater when nitrogen fertilizer was applied to simulate an environment with greater nitrogen supply (77% greater number of volunteer B. napus seeds produced with vs. without broadcast application of 23 kg N ha−1 urea). In northern climates, seeding soybean at increased densities using narrow-row spacing in fields with limited soil inorganic nitrogen and using interrow tillage in wide-row production systems are effective strategies that could augment chemical weed management in an integrated program for management of volunteer B. napus, and perhaps also other competitive early-season weeds.

In west central Minnesota the extent and duration of weed seed shed was measured for two years in corn that received cultivation but no herbicides. Percentage of seed production represented by viable (filled) seeds was about 79% for green foxtail, 68% for wild mustard, 49% for Pennsylvania smartweed, 48% for common lambsquarters, and 35% for redroot pigweed. Percentage viable seeds varied from 11% in 1993 to 59% in 1994 for redroot pigweed, but was more stable for other species. Seed shed commenced in late August in a cool year (1993) and early August in a warm year (1994). Average growing degree days (base 10 C) from corn planting until 25% seed shed was 983 for common lambsquarters, 984 for wild mustard, 1004 for Pennsylvania smartweed, and 1034 for both green foxtail and redroot pigweed. Brief weather events, such as wind storms, dispersed large percentages of total seed production within a single day. More than one-fifth of all viable seeds of green foxtail, redroot pigweed, and common lambsquarters were retained by the seedheads and dispersed by combines at harvest. In contrast, seeds of early-maturing species, such as wild mustard, were completely dispersed before corn harvest in the warmer year, but one-third of seeds were retained by seedheads at harvest in the cooler year. Measurement of seed shed was compared using five seed trap designs. The preferred design consisted of a 10-cm-diam plastic cup, whose bottom was replaced by a brass screen, and the entire unit attached to a small wooden stake for support. This design provided, on average, the highest estimates of seed production, least among-replication variability, highest correlation with weed population density and aboveground dry-weight, lowest assembly cost, and greatest ease for sample access and seed processing.

The efficacy of current wild oat herbicides and their high cost have resulted in use rates that are less than those recommended. While acceptable weed control may be attained at less cost, this practice does not consider the potential for increased wild oat seed production. The objective of this experiment was to determine the interaction of wild oat density and reduced imazamethabenz rates on wild oat seed production in spring barley. As wild oat densities increased from 8 to 1,100 plants m−2, wild oat seed production increased from 180 to 9,950 seed m−2 without herbicide, and from 0 to 2,810 seed m−2 using 0.53 kg ai ha−1 imazamethabenz. This general pattern was modeled using a cumulative logistic function. Estimates from this model indicated that < 1 wild oat seed m−2 was produced at population densities of ≤ 20 plants at any imazamethabenz rate. Imazamethabenz rates of 0.26 kg ha−1 or greater at wild oat densities of less than approximately 190 plants m−2 did not result in wild oat seed production above the initial population density. As wild oat density increased, however, imazamethabenz rates below 0.40 kg ha−1 resulted in substantially greater wild oat seed production compared to the recommended rate.

Studies were conducted in conventional and no-tillage corn in 1990, 1991, and 1992 at Wooster, OH, to measure corn yield and velvetleaf seed production in response to density of early and late emerging velvetleaf, and to estimate economic thresholds. The percent reduction in corn yield fit a hyperbolic function over velvetleaf densities from 1 to 30 plants m2. The percent yield loss and velvetleaf seed production were higher in a warm, wet year (1990) than in a dry (1991) or cold, wet year (1992). The percent corn yield reduction was generally greater in no-tillage than in conventional tillage and from early rather than late emerging velvetleaf. Maximum velvetleaf seed production ranged from about 18,000 seeds m2 for early emerging weeds in no-tillage in 1990 to 100 seeds m2 for late emerging weeds in no-tillage in 1992. The single year economic threshold for early emerging velvetleaf ranged from 0.40 to 14.0 velvetleaf m2 in conventional tillage and 0.13 to 3.13 in no-tillage. Economic thresholds that were predicted using yield goal information deviated from actual thresholds (using actual yields) for a given year by −43 to 30%. Single year economic thresholds were similar in both tillage treatments, but their value for management decisions is questionable due to variation among growing seasons and weed seed production from subthreshold populations.

Studies were conducted to determine the effect of interference between ladysthumb and cotton on plant growth and productivity. Ladysthumb remained shorter than cotton until at least 70 d after cotton planting. However, ladysthumb grew over twice as tall as cotton and, depending on plant density, produced between 179 and 681 g dry biomass per plant by cotton harvest. Ladysthumb biomass per plant was not affected by weed density when grown with cotton. When grown alone, ladysthumb produced over 2,000 g dry biomass per plant, which was over four times greater than biomass produced by plants grown with cotton. Cotton lint yield decreased between 0.7 and 0.9 kg ha−1 with each gram increase in weed dry biomass per meter of the row. The relationship between ladysthumb density and cotton percent yield loss was described by the rectangular hyperbola model with the asymptote (coefficient a) constrained to 100% maximum yield loss. The estimated coefficient i (yield loss per unit density as density approaches zero) was 35 ± 5 and 14 ± 2 in 1998 and 2000, respectively. Ladysthumb seed production was also described by the hyperbolic function. Estimated seed production at 1 plant m−1 of cotton row was 33,000 and 47,000 seed m−2 in 1998 and 2000, respectively.

Seed production of weedy species of Setaria in crops of Zea mays and Glycine max was studied for 2 yr in western Minnesota and eastern South Dakota. Viable seed production was curvilinearly related to panicle length. A 100-mm-long panicle of S. pumila, S. faberi, and S. viridis produced 129, 323, and 851 viable seeds, respectively. Values were consistent across years, crops, and herbicide treatments. Frequency distributions of panicle lengths of all panicles within a population closely followed nonlinear Weibull functions and were stable across years and crops but not species or herbicide treatment. Positive skewness of these distributions decreased, and median panicle size (mm) increased, in the following order: S. viridis (41), S. pumila (52), and S. faberi (78). Postemergence herbicides applied at full label rates increased skewness and reduced median panicle size (to 11 mm) and seed production of S. viridis. Skewness lessens the reliability of using average panicle size as a measure of seed production for the entire population. However, integration of panicle size–frequency and panicle size–fecundity relationships provided estimates of the number of seeds per panicle that were more representative of the population than the statistical average panicle. These estimates were 52, 242, and 246 seeds per panicle for S. pumila, S. viridis, and S. faberi, respectively. Multiplication of these values by panicle densities generated seed production estimates that were similar to actual counts of seeds. Setaria seed production tended to be higher in Z. mays than in G. max only because of higher plant and panicle densities. Early-maturing panicles tended to be larger than those maturing later, but seed viability generally was stable across maturity times.

Winter annual weeds protect the soil from erosion and retain nutrients during the winter; however, they can also act as a host for crop pests and pathogens and impede planting. Increased knowledge of the reproductive biology and the seed fate of winter annuals would be useful to improve management and crop productivity. The objectives of this research were to determine the recruitment biology of shepherd's-purse, henbit, common chickweed, and field pennycress, including seed production, dispersal, dormancy, and seedling emergence, based on growing degree days (GDD). Henbit was the least prolific of the four weeds studied, producing 800 to 40,000 seeds m−2 at naturally occurring densities; shepherd's-purse was the most prolific, producing 11,000 to 400,000 seeds m−2 with 40 to 230 plants m−2. Fifty percent seed rain occurred for henbit, common chickweed, shepherd's-purse, and field pennycress at 620, 790, 880, and 1300 GDDBase ,0C, respectively. Overall, seeds were dormant for all species at the time of dispersal. In 2 of 3 yr, dormancy of later-dispersed common chickweed decreased after 6 mo of storage at natural, fluctuating temperatures in the absence of water. The emergence patterns of the four species followed the Gompertz equation and were indicative of facultative winter annuals. The emergence patterns by rate were similar between henbit and common chickweed and between shepherd's-purse and field pennycress. Seed production, dispersal, dormancy, and seedling emergence were influenced by moisture; therefore, including a precipitation or soil moisture component into a GDD model (such as the use of hydrothermal time) would improve the accuracy of predicting winter annual reproduction, seed fate, and emergence.

Effective, long-term wild oat management requires an integrated approach that uses management techniques beyond simple herbicide application. A 5-yr (1996 to 2000) zero tillage study was conducted to assess the influence of barley harvest timing on wild oat densities in subsequent years at Lacombe, AB, Canada and Melfort, SK, Canada. Harvest timings included barley harvested 1 wk after heading (early), approximately 14 to 16 d later at the soft dough stage (normal), and at maturity (grain). In the absence of herbicides, wild oat densities decreased in silage plots harvested early and increased in grain plots. Reductions were more distinct at Lacombe where barley phenological differences and whole plant moisture contents between early and normal silage harvests were greater than at Melfort. Half rates of wild oat herbicides (ICIA 0604 and imazamethabenz) did not augment reductions in wild oat densities after early silage harvest, but did improve wild oat management after normal silage harvest, and in grain production. At Lacombe, early silage harvest reduced wild oat densities more than did herbicides in grain production. Similar trends were apparent at Melfort but were not statistically significant. Early barley silage harvests may be an effective integrated weed management tool for wild oat.

Relationships among weed seed rain, soil seedbank, and seedling recruitment in no-tillage systems were studied from July 1993 to May 1996. Multiple regression analysis indicated that seedling recruitment of only six of the 25 weed species present was correlated with seed rain samples from the previous autumn, spring soil seedbank samples, or a combination of the two. However, seedling recruitment of the dominant annual grasses (yellow foxtail, giant foxtail, and fall panicum in Field 1–1994, Field 2–1995, and Field 3–1996, respectively) was related to seedbank populations or a combination of seedbank and seed rain densities. These grasses accounted for at least 32% of the emerged seedlings, 12 to 78% of the seedbank, and 16 to 77% of the seed rain. Seedling recruitment of large crabgrass and two broadleaf species, Virginia copperleaf and wild carrot, also were described by seedbank densities or a combination of seedbank and seed rain densities. However, both the broadleaf species were minor components of the cropping system, representing ≤ 4% of all seedlings. In each year, the sum of all weeds in the seedbank exceeded 1,300 seeds m−2 (0 to 10 cm deep). The fraction of the total seedbank that emerged each year ranged from 3 to 17%, but there was great variability among species. The aforementioned dominant grasses ranged in emergence from 3 to 25% of their seedbanks. Emergence of other species ranged from a low of 1% for common lambsquarters up to 41% for large crabgrass. Prediction of seedling recruitment from seed rain or seedbank densities was variable; however, combining both components improved the fit of regression describing seedling recruitment.

Avena fatua seeds remaining on the plant at harvest and taken into the combine harvester may be dispersed over large areas. The objective of this study was to characterize the development of A. fatua in comparison to spring Triticum aestivum. As part of this objective, the rate of seed shed in A. fatua relative to development of T. aestivum was determined. Avena fatua and T. aestivum had similar phyllochron intervals within locations but differed between locations. Plant development as measured by the Zadoks plant development scale was consistent within plant species between locations. Seed shed in A. fatua was also consistent between locations. Most of the seed shed occurred within 2 wk, and the cumulative seed shed followed a sigmoidal pattern. The seed shed occurred as T. aestivum was ripening, and the percentage of seed shed appears to be related to the water content of the T. aestivum spike. Because of this relationship, the proportion of seed remaining on A. fatua at harvest could be managed by changing the timing of crop harvest.

At the time of grain harvest, weed seeds can be classed into one of four pools on the basis of dispersal status and location: (1) undispersed, remaining on the mother plant; (2) dispersed in the current year, on the soil surface; (3) dispersed in the current year and collected by harvest machinery; and (4) dispersed in a previous year and persisting within the soil seed bank. Knowledge of the relative sizes of these seed pools for different weed species under different crop environments will be useful for determining the best way to reduce the size of inputs to the soil seed bank. In fall 2004 and fall 2005, four randomly selected commercially managed corn and soybean fields in east-central Illinois were sampled to quantify weed seed pools at time of crop harvest. Thirty randomly located 0.125-m2 quadrats were placed within each field, the four seed pools mentioned above were sampled for each quadrat, and the species composition and abundance of each seed pool was determined. The magnitude of the weed seed rain varied among species and between years and crops. Twenty-six weed species were found to contribute to at least one of the four seed pools. However, the weed seed pools were consistently dominated by six species: velvetleaf, Amaranthus complex (redroot pigweed and waterhemp), ivyleaf morningglory, giant foxtail, prickly sida, and common cocklebur. For each of these species, the ratio of undispersed seeds to seeds in the soil seed bank at harvest time was ≥ 1 in at least one crop during one of the two experimental years, indicating a potential for the soil seed bank to be completely replenished or augmented by that year's seed rain. This analysis demonstrates the urgent need for techniques to limit weed seed inputs to the soil seed bank at the end of the growing season.

Populations of plumeless thistle and musk thistle were mowed at various growth stages at two separate sites in Maryland during a 6-yr period to elucidate relationships among seed rain, soil seed banks, and population recruitment. The majority of seeds (96%) in the soil profile were distributed within 7.6 cm of the surface at both sites. Mowing plumeless thistle when most of the flower heads were at the full bud or postbloom stage did not reduce seed bank or plant densities, unlike mowing at full bloom, which significantly and quickly reduced both. Musk thistle responded differently, with plant density declining only with the postbloom mowing treatment, which occurred after the parent plant had dispersed seeds and died. Seed bank densities were unchanged by this treatment. This indicates that other factors may limit the recruitment and maintenance of musk thistle, such as allelochemical production by parents and interspecific plant competition. Density of musk thistle declined over an 11-yr period at one monitoring site, whereas plumeless thistle remained unchanged. The disproportionate seed-destroying activities of the weed biological control agent Rhinocyllus conicus on musk thistle may explain this difference.

Studies were conducted to determine the effect of interference between Pennsylvania smartweed and cotton on plant growth and productivity. Pennsylvania smartweed remained shorter than cotton until at least 80 d after cotton planting. However, Pennsylvania smartweed produced considerable dry biomass by cotton harvest. Pennsylvania smartweed biomass per plant was not affected by weed density when grown with cotton. When grown alone, Pennsylvania smartweed produced 1,640 and 2,060 g dry biomass plant−1 depending on the year. This biomass was over four times greater than the biomass produced by plants grown with cotton. Cotton lint yield decreased between 1.3 and 1.1 kg ha−1 with each gram increase in weed dry biomass per meter of row. The relationship between Pennsylvania smartweed density and cotton percent yield loss was described by the hyperbolic function. The estimated coefficients a (maximum yield loss as density approaches infinity) and i (yield loss per unit density as density approaches zero) were 102 ± 23 and 51 ± 12, respectively, in 1998 and 53 ± 1 and 98 ± 5, respectively, in 2000. Pennsylvania smartweed achene production was also described by the hyperbolic function. Estimated achene production at 1 plant m−1 cotton row was 18,000 and 26,000 achenes m−2 in 1998 and 2000, respectively.

Field studies were conducted at two North Carolina locations to determine the effect of interference between pale smartweed and cotton on plant growth and productivity. Pale smartweed remained shorter than cotton until at least 70 d after cotton planting. However, pale smartweed grew over twice as tall as cotton and produced considerable dry biomass by cotton harvest. Pale smartweed biomass per plant was not affected by weed density up to 3.5 plants m−1 of row when grown with cotton. Cotton competition reduced pale smartweed dry biomass per plant at least 400%. The relationship between pale smartweed and cotton percent yield loss was described by the rectangular hyperbola model with the asymptote (coefficient a) constrained to 100% maximum yield loss. The estimated coefficient i (yield loss per unit density as density approaches zero) was 29 ± 4 and 23 ± 4 in 1998 and 2000, respectively. Pale smartweed achene production was also described by the hyperbolic function. Estimated achene production of smartweed at 1 plant m−1 cotton row was 63,000 and 25,000 achenes m−2 in 1998 and 2000, respectively.

Seed-dispersal ecology in tropical montane forests (TMF) differs in some predictable ways from tropical lowland forests (TLF). Environmental, biogeographic and biotic factors together shape dispersal syndromes which in turn influence forest structure and community composition. Data on diaspore traits along five elevational gradients from forests in Thailand, the Philippines, Tanzania, Malawi and Nigeria showed that diaspore size decreases with increasing altitude, fleshy fruits remain the most common fruit type but the relative proportion of wind-dispersed diaspores increases with altitude. Probably corresponding to diaspore size decreasing with increasing elevation, we also provide evidence that avian body size and gape width decrease with increasing altitude. Among other notable changes in the frugivorous fauna across elevational gradients, we found quantitative evidence illustrating that the proportion of bird versus mammalian frugivores increases with altitude, while TMF primates decrease in diversity and density, and switch diets to include less fruit and more leaf proportionately. A paucity of studies on dispersal distance and seed shadows, the dispersal/predation balance and density-dependent mortality thwart much-needed conclusive comparisons of seed dispersal ecology between TMF and TLF, especially from understudied Asian forests. We examine the available evidence, reveal knowledge gaps and recommend research to enhance our understanding of seed dispersal ecology in tropical forests. This review demonstrates that seed dispersal is a more deterministic and important process in tropical montane forests than has been previously appreciated.

Nucleation, leading to the formation of tropical forest patches in open areas, has occurred in many parts of the world. This study examined the role of Ficus natalensis as a nucleus and estimated seed dispersers in the nucleation process. Seed rain and post-dispersal fate of Syzygium guineense ssp. afromontanum (a common forest tree species) under eight F. natalensis crowns were compared with those in other woodland microsites. In addition, focal observations of frugivore were conducted at F. natalensis in the woodlands and S. guineense ssp. afromontanum in the forest, when both species were fruiting. Most dispersed seeds (85%) were found under F. natalensis, and the number of dispersed seeds was significantly greater under F. natalensis than at other microsites. Germination (n = 600) and survival percentages (n = 384) were greater under F. natalensis compared with those in treeless open microsites. These results suggest that F. natalensis act as efficient nuclei in nucleation process. Results of focal observations at S. guineense ssp. afromontanum (100 h) showed that Schalow's turaco consumed 63% of fruit and swallowed all pecked fruit. Schalow's turaco also visited fruiting F. natalensis frequently (0.4 h−1) during 80 h of observations, indicating that Schalow's turaco is an important seed disperser during nucleation.

Survival of Afrotropical primary forests depends not only on habitat protection but also on the protection of animal species such as frugivorous primates, recognized as the most important seed dispersers for many plants. Here we investigate seed-dispersal services by the bonobo (Pan paniscus) in an evergreen lowland tropical rain forest of the Congo Basin. In the long-term research site of LuiKotale, we investigated food habits and seed processing based on 22 mo of behavioural observation, seed trial experiment and long-term daily GPS tracking of a habituated ape community. Bonobos were mainly frugivores (66% of all feeding sessions), spending about 3.5 h d−1 swallowing seeds that were transported for an average of 24 h in the gut. On average, an individual bonobo dispersed 172 kg y−1 of seeds (or 220000 seeds y−1) of more than 91 plant species by endozoochory over an average distance of 1.2 km from the parent tree. Passed seeds germinated more rapidly, more successfully and had greater post-dispersal survival than unpassed seeds. Bonobo-dispersed plants accounted for 40% of tree species and 65% of individual trees in the study site (12 1-ha plots census). Since bonobos show little functional overlap with other frugivores, the loss of their seed-dispersal services is likely to affect forest structure and dynamics. Our results justify description of the threatened bonobo as a gardener of the Congo forest.