Ramp Research Summary, 2007-Present
In 1993 the Journal of Ecology published a study by Nault and Gagnon, titled "Ramet demography of Allium tricoccum, a spring ephemeral, perennial forest herb," which you can read here https://www.jstor.org/stable/2261228. The authors recorded population growth rates of ramps in dense populations. They then created a model simulating harvest, assuming that population growth rates after harvest were the same as those in unharvested dense populations. Since this model ignored a basic principle of population ecology--namely, density-dependent mortality effects, or competition--its results were fundamentally flawed. Basic ecological tenets posit that plants at low densities experiences less competition that plants at higher densities (under like growing conditions), and thus the lower density plants have access to more resources, which means they grow and reproduce faster and have lower mortality rates. Despite ignoring this basic principle, the Nault and Gagnon study was highly influential, and has been the primary basis for the vehement anti-foraging ramp-shaming that has become common over the past decades. Although we fully support responsible harvest of all wild foods, including ramps, this ramp-shaming is part of an anti-foraging movement and attitude that threatens ancient food traditions, and the plants themsleves. Conservation is prompted by appreciation of Nature, which is fostered and strengthened by traditions steeped in wild plants, animals, fungi, and landscapes. We believe that harvest of ramps is a vital aspect to the long-term conservation of this species and the deciduous forests in which it thrives. In response to questions about the unsustainability of ramp harvest, we started a long-term research project in 2007 to study numerous aspects of ramp ecology, to better understand sustainable harvest and management practices. Particularly, we wanted to look at the reproduction rates of ramps at lowered population densities, such as would occur after harvest. We also wanted to understand the pant's ecology, to better work with it to design and implement harvest and management practices. On this page I will be summarizing the results of these various research projects. More detailed write-ups and analysis will eventually be published in peer-reviewed journals, but I wanted to make this data available in the meantime, for those people interested in realistically understanding ramp ecology and demography.
This page is a work in progress. Please be patient. I have put thousands of hours into this, and thousands of dollars of my own money. Nobody is paying me for this. I have received no grants or other financial support from anyone. I have a busy life and will be updating and expanding the data here as time allows. Eventually I will probably set up an entire website for this research, after which this page will redirect readers there. Please email me with suggestions, concerns, ideas, etc.
Our research is primarily set in the Town of Hill, Price County, Wisconsin, at 45.5 N, 90.2 W, and approximately 550 meters of elevation, at our active, small, commercial maple sugaring operation. The terrain is terminal moraine with steep slopes in all aspects. The soil is glacial till with a moderate amount of clay and many large stones. The forest is dominated by sugar maple Acer saccharum (about 60% and increasing with management), white ash Fraxinus americana, basswood Tilia americana, American elm Ulmus americana, and yellow birch Betula allegheniansis, in respective order of dominance. Other trees present in small amounts include black cherry Prunus serotina, pin cherry Prunus pensylvanica, northern red oak Quercus rubra, red maple Acer rubrum, white bich Betula papyrifera, hophornbeam Ostrya virginiana, black ash Fraxinus nigra, eastern hemlock Tsuga canadensis, butternut Juglans cinerea, and slippery elm Ulmus rubra. The site was northern hemlock-hardwood forest with scattered white pine at the time of European settlement. The hemlock was removed in the early 1900s. The residual and second-growth hardwood stand was logged heavily about 1965, and most canopy trees in 2020 are 55-90 years old. This site is lucky to have adjacent old-growth stands for comparison, with most canopy trees in 240-315 years of age (as determined by growth ring counts on fallen trees). In these old-growth stands hemlock is the most dominant tree, and under the hemlocks no ramps or spring ephemerals are present. In some nearby hardwood stands with a dense ground layer of Canada yew Taxus canadensis the spring ephemeral herbs are also absent. The study site contains several small, robust patches of ramps and other spring ephemerals, such as Virginia waterleaf Hydrophyllum virginianum, Common spring beauty Claytonia virginica, and cut-leaf toothwort Cardamine concatenata, but is low in diversity and poorly stocked with typical herbs of a mesic northern hardwood forest. The few colonies of mesic forest herbs tend to be localized but large and nearly monocultural--not highly integrated with other species. This suggests recent and active colonization by these species, and means that most of the forest is ideal habitat for ramps, with very little competition, presumably due to the past dominance of hemlock and presenve of Canada yew precluding the spring ephemerals.
Basic Biology of Ramps, AKA Wild Leeks, (Allium tricoccum and Allium burdickii)
The taxonomy of ramps in eastern North America is contentious or complicated. For many years a single species, Allium tricoccum, was recognized. In 1979 this was split into two species, with the less common segregate being designated Allium burdickii. The latter species, sometimes called "white ramp," "green ramp," "small ramp," or "narrow-leaf ramp," is differentiated by being smaller, having narrower leaves, having a green (as opposed to reddish) leaf sheath, blooming about a month earlier, with scapes emerging well before leaf senescence. The umbels of A. burdickii are smaller and contain fewer flowers and seeds. The onion/garlic scent of its leaves is also weaker than that of the typical "red ramp" or "large ramp", ie, A. tricoccum. The common name "ramps" is a corruption of the English name "ramson" applied to a very similar and closely related mesic forest species of Europe.
In many locations the two North American ramp species are clearly differentiated. However, there are also extensive populations in which the characteristics that supposedly separate them are mixed. Some of these populations are hard to assign to either of the described species. Some researchers are currently looking at whether or not there are more than two species of ramps. This is not likely to be resolved by genetic work, however, because the definition and boundaries of species are probably the most problematic question in all of biology, and conclusions reside in the confluence of philosophy and fact. My personal working hypothesis is that we have two species of ramps that differentiated through past separation, but which have now come back into contact. Their processes of genetic isolation appear to be incomplete, perhaps more so in the northern part of their distributions, and currently it seems that they are re-converging, and some populations appear to be of hybrid origin. So, whether we consider them one species or several, suffice it to say that ramps are highly variable in size, phenology, and leaf characteristics, and their taxonomy is not settled. (And evolutionary theory implies that, in many cases, taxonomy should not be settled.)
Ramps are deciduous forest spring ephemerals; they emerge in early spring to take advantage of the brief growing period during which the canopy trees are devoid of leaves, and ample light hits the forest floor. Soon after the leaves appear on the overhead deciduous canopy, the ramp leaves senesce (die back) for the season. This makes for a typical growing season of only 4-9 weeks. However, scape production, flowering, seed production, and bulb division occur after the leaves have senesced, all relying on energy stored in the bulb. Flowering typically occurs in early to mid summer, and seeds ripen in late summer, although they may persist well into the fall, and even into the following summer. Seeds are dispersed by gravity, water, rodents, and some birds. The seeds do not germinate in their first spring; they germiante the second spring, about 19 months after maturing. Seedlings require at least seven years to reach sexual maturity (six years of growth, the first year being dormant). Bulb division occurs mostly in the late summer after the production of a scape; occasionally it occurs without scape production. Scape production rarely occurs unless a certain biomass is attained. After scape production, bulbs shrink or divide; they may flower in consecutive years, but more often store energy through one or more growing seasons before flowering again.
Ramps are found on many soil types of moderate to high fertility: clay, clay loam, loam, alluvial deposits of sand and silt mixtures, ground moraine, and glacial till. They are absent from very poor or thin soils, or those with very low capacity to retain moisture. They need neutral to slightly alkaline soils, and seem to thrive best where calcium levels are high. The classic tree associate is sugar maple, but ramps also thrive under basswood, red maple, siver maple, elms, yellowbud hickory, ash, yellow birch, buckeye, hornbeams (Carpinus and Ostrya), mulberry, butternut, balsam poplar, and many other species. Allium burdickii appears to thrive better in heavy clay soils. Allium tricoccum seems to grow largest on soils derived from limestone, or Appalachian slopes.
Definitions:
Ramet: This refers to a separate, independent biological unit of an organism that can reproduce asexually. Technically, each bulb in a clump of ramps may be genetically identical, though physically separate. The separate bulbs could be thought of as one organism broken into pieces, or ramets. Since there is no way to tell if a mature pre-existing bulb was produced by seed or bulb division, we call each bulb and its associated structures a ramet.
Study 1:
Population Growth and Demographics of Ramps Allium tricoccum at Low Densities: Analysis of Growth from Founder Bulbs Transplanted in 2007.
In this study we transplanted 85 ramp bulbs (each of which we call a founder bulb) to an area of suitable northern hardwood forest from which ramps were absent, so that we could monitor the population increase without the confounding variable of pre-existing plants. On May [ ], as the ramp leaves had reached full size, but before they had begun to senesce (die back) for the season, the bulbs were carefully dug out of one colony and immediately transplanted 30-60 meters away in two rows running up a rather steep south-facing hillside. The founder bulbs were placed approximately [ ] cm apart (although the distance was often not uniform due to obstructions such as logs and rocks). A small shovel was inserted about 6 inches (15 cm) into the soil and pulled back to create an opening into which the bulb was inserted about half way, allowing the roots to extend to the bottom of the hole. The soil was replaced around the bulb and roots, patted down, and less than half a liter of water was administered. This operation took less than one minute per transplant; after this, no care was given to the transplants. In 2012 the top broke off a white ash tree and landed in the west row—this top had a severe negative impact on five of the colonies but was allowed to decay in place rather than being removed. Although this study took place in an active commercial sugarbush, no thinning or other cutting of trees was done in the study area from 2007 to 2022.
In 2017 we assessed survival and reproduction. For each colony (defined here as the plants surrounding a founder bulb) we recorded each ramet (bulb), its number of leaves, and the width of each leaf at the widest point (to the nearest mm). We assigned each ramet as being derived from direct vegetative reproduction (bulb division), seed reproduction, or bulb division of a seedling. These categories were marked in our data as V, S, and VS respectively. The founder bulb was classified as V. (Although in some cases it was not possible to tell if the original founder bulb had died and was replaced by one of its daughter bulbs.) We also recorded the widest visible spread of seedlings from its apparent founder bulb.
There were a few uncertainties in our data collection. One is that it is impossible to know the seed source for any particular seedling—rain or gravity could have moved seeds between founder bulbs. This made it impossible to determine the greatest downhill spread of seedlings—an important weakness, because gravity should effect a greater spread downhill than in other directions. Thus, seed spread from founder bulbs was likely significantly underestimated. However, such movement within the plot would have had no effect on the totals for the study plot as a whole, and thus no impact on the calculated overall reproduction rate or demographics.
A second possibility is that seeds could have been transported into or out of the study plot by mammals or birds, as we know that one or more species consumes and transports the seeds (See Observations). However, since the density of ramps inside the study plot is greater than the overall average density of ramps on the surrounding landscape, both on an intermediate scale (5-50 meters, as is likely to be relevant for rodents), and on a broader scale (100-1000 meters, as is likely to be relevant for birds), the probability is that more seeds would have been transported out of the study area than into it. Thus there is reason to suspect that the observed reproduction may be slightly underestimated by outgoing seeds, and no reason to believe that it is inflated by incoming seeds.
Finally, a seedling that germinates close to the founder bulb can be mistaken for a daughter bulb. First-year daughter bulbs are much larger on average than seedlings at first emergence, and often have two leaves (as opposed to the single leaf always produced by new seedlings). Even when daughter bulbs have one leaf only, these are usually proportionately much broader than single seedling leaves. However, after 3-5 years these differences between daughter bulbs and seedlings are no longer evident. Since we did not collected data until 10 years after transplanting, it is probable that a small portion of the ramets designated as daughter bulbs were actually seedlings. Because of this uncertainty, we chose to designate all ramets with any bulb portion located within 2 cm of the founder bulb or founder clump (meaning the clump of ramets that arose vegetatively from the founder) as vegetatively reproduced. This means that, almost certainly, our data slightly inflates the percentage of reproduction allocated to direct vegetative reproduction, and underestimates the percentage allocated to seed.
Demographic data was collected again in 2018, 2020, and 2022. As the colonies expanded, the boundaries of the seedling zones around each founder became progressively harder to determine, so in the 2022 data collection, we tallied the lump sum of the ramets in each row, but no longer assigned ramets to any particular founder bulb or colony.
From our original 85 bulbs (ramets), the following totals were regenerated:
as of 2017:
2018: 1,324 ramets
2020: 1,937 ramets
2022: 2,851 ramets
Please allow time for further analysis. I have hundreds of pages of data for this study alone. I will break down the population by size class (leaf number), and also total leaf width (TLW), which is used as a proxy measure for biomass without having to kill the plants to weigh them. Since the original founder bulbs were all mature and had at least two leaves, while the later ramet tallies include a significant percentage of smaller seedlings, the percent increase of the TLW index should be somewhat lower than the figures for ramet number increase.
Another very important question that our data will address is the percentage of reproduction resulting from seed versus bulb division. I will post this as soon as I get the relevant totals.
Discussion
In 15 years, this is an increase of 33.54 times the original count, or a 3,354% increase. Please compare this to the increase rates modeled in Nault and Gagnon (1993). Note that Nault and Gagnon did not observe the population declines, they just predicted them, using a model designed to predict them, which could reasonably have no other outcome. Please note that these increases are based on very low initial population densities. Responsible harvesters would not reduce the population to the levels at which our study began, so the rate of increase should be lower in a harvested natural stand of similar quality.
Study 2:
Transplant Success at Different growth Phases, and the Effects of Competition from Pennsylvania Sedge Carex pennsylvanica on the Survival and Reproduction of Transplanted Ramps Allium tricoccum.
This study was designed with two purposes. The first was to test the relative success of transplanting ramp bulbs at 3 different growth stages: early emergence (leaves less than 3 inches long; this stage usually occurs at the study site for a brief period in April or early May), peak growth (leaves at full size, but before senescence begins; usually mid May), and summer dormancy (after the leaves have completely senesced). I established 3 plots, each divided into 3 rows of 15 bulbs each, for a total of 45 bulbs per plot. A few large obstructions, such as branches or rocks, were removed from each plot. In each plot, one row of bulbs was planted in each of the three growth stages, making for a total of 45 bulbs transplanted in each growth stage. The growth stage planted in each row was assigned randomly. Transplants for all 3 growth phases were pulled from the same nearby wild colony and sorted according to this protocol: Any daughter bulbs were separated, after which bulbs were selected for transplanting only if they were undamaged and had at least 6 roots 2 cm or more in length. After a large pool of qualifying bulbs was amassed in a tub, they were selected blindly, one at a time, by grabbing the first bulb or ramet to contact my fingers, and then shaking the tub between selections. Selected bulbs were placed in 3 separate piles of 15, after which a random numbers table was used to determine which pile was planted in which plot, as well as the order of planting the plots. The selection and transplanting for each growth phase was completed in one day, as rapidly as possible. A hole about 6 inches (15 cm) deep was opened with a trowel, a bulb was inserted with the lower terminus about 3 inches (8 cm) below the surface so the roots could rest in a natural position, and the soil was immediately pressed back into place over the bulb. This took about 20 seconds per bulb. No further care took place at any time.
While transplanting success or failure was apparent after two growing seasons, the study was designed to simultaneously test the long-term effects of competition from Pennsylvania sedge Carex pensylvanica. In two of the plots Pennsylvania sedge was absent or in very low density; the third plot was characterized by a dense covering of Pennsylvania sedge over the entire area. While transplanting success is important to recovery, management, and ecoculture practices involving ramps, the impacts of interspecific competition are also extremely important for such projects, and are even more important to understanding the ecology of this species in spontaneous wild populations.
Study 3:
Scape production rates among populations of ramps Allium tricoccum at high and low densities.
This study was designed to understand how the density of ramets in a population affects the production of scapes (flowering stems). Specifically, we are interested in whether or not the harvest of bulbs results in an increase in the scape production rate (and thus, reproduction and recovery) of remaining plants, as is predicted by the basic tenets of population ecology (Tilman, 1987; ); and contrary to the predictions of Nault and Gagnon (1993), who suggested that residual populations after harvest would show no increase in reproduction, and indeed might show increased mortality. The data was collected in late spring of 2022 (June 1), two growing seasons after bulbs had been harvested from some of the plots for transplanting. Unlike some of the projects, long-term monitoring for multiple years is not necessary to answer this question. However, additional sampling, as time permits, will strengthen the conclusions, and such data can be added at any future date, through replication of the sampling protocols we used, or alternative protocols which might have advantages.
We sampled in two separate ways. First, we placed quadrats of one meter squared, built from one-inch PVC pipe, into colonies of ramps, so that no portion of the quadrat was more than 30 cm from an established bulb. No quadrat placement overlapped any other previous placement. We sampled all of the colonies sufficiently large for quadrat placement within a designated section of our sugarbush; one colony was large enough to allow two quadrats. In total, 11 quadrats were sampled. A ramet was included in the quadrat if the upper terminus of its bulb was located within the quadrat; leaves leaning in from outside ramets were not counted. Bulbs directly under the quadrat border were included or excluded based on the natural location of their leaves.
Two of the sampled quadrats were in colonies established by seeding a decade or more prior to sampling, from which no ramets have been harvested. The other 9 quadrats were placed within spontaneously occurring wild colonies, 8 of which had been subject to some level of harvesting 2 years prior to sampling.
We counted each ramet in the quadrat, recording the number of leaves, and if a scape or scape bud was present. At this time some scapes were evident as buds in the juncture of the leaves, while other scapes had extended up to 14 centimeters. Although we checked systematically for scape buds, it is likely that we missed a few. However, past data collection indicates that nearly all scapes are evident at this stage of growth, and there is no reason to suspect a systematic bias in scape appearance between low and high density populations, so these few missed scape buds should have no material effect on the accuracy of the data. We did not record leaf width because 1) Some of the leaves were beginning to senesce, making accurate width measurement impossible, and 2) Although leaf width would add some value to the data, it was immaterial to the basic analysis we were aiming for.
Our sampling was for scapes only. We have consistently observed a very strong positive correlation between scape production and vegetative reproduction, and this relationship is borne out by the data of other researchers in all cases where it has been examined (Nault and Gagnon, 1993). We assume that there is also a positive correlation between scape production, flower production, and seed production. However, it is probable that this relationship is not linear, and our observations strongly suggest this. Scapes in lower-density populations appear to be larger on average, and to produce more flowers and seeds. Furthermore, due to seed predation, and the tendency of larger populations to attract vertebrate seed predators, or to harbor populations of insect seed predators, there also appears to be a complex relationship between scape density and successful production of viable seeds (see notes on study 2). Although the current study provides valuable insight into density-dependent reproduction of ramps, we intend to design and conduct research in the future to better understand potential differences in seed production per scape.
There is an important practical reason that we sampled for scapes only rather than flowers or seeds. Allium tricoccum is one of the few herbaceous species in our region that flowers when the leafy parts are dormant and thus not available for examination. If we were to count actual flowers or seeds, there would be no leaves available at the time of data collection, so we would not be able to effectively determine the density of non-flowering ramets in a plot, nor the leaf number, or leaf width, of these ramets. Our protocol allowed us to collect all of the data at once.
The obvious solution to this dilemma is to leave physical quadrats in place through the entire growing season, and collect the data at multiple times. This would allow us to see not only what percentage of ramets produce scapes, but also how many flowers and seeds were produced per scape. We plan to do this in the future, when we find the time to do it, perhaps at different study sites closer to home.
To the above 11 quadrats we add the 2022 data from two long-term quadrats in which we are monitoring ramet count, leaf count per ramet, and leaf width, as well as scape production (see study 4). These quadrats have seen no bulb harvesting; one of them has had leaves harvested.
The second means of data collection involved tracking the transplanted bulbs removed in 2020 from some of the colonies sampled by quadrat. These were transplanted to another section of the sugarbush with appropriate tree cover and soil, but totally devoid of ramps (and mostly devoid of spring ephemerals). They were transplanted on a rough grid, 50-70 cm apart, by opening a small hole about 15 cm into the soil with a trowel, holding a bulb with the bottom about 8 cm deep in the hole, and replacing the soil. This procedure took 15-20 seconds per bulb. No care was administered after transplanting.
We sampled this area by going down the rows of transplants and recording the number of ramets, their leaf counts, and the presence of scapes. Because of landscape idiosyncrasies that resulted in disorganization of the grid, we certainly missed many of the transplants, but there was no systematic bias that would threaten the reliability of the data. We did not calculate a density per meter squared for this area, but rather considered it “extremely low density, functionally equivalent to no intraspecific competition.” An average density per square meter in the transplant area would probably come to 4-7 ramets. However, in the future we will collect stronger and more readily comparable density data by marking the boundaries of the sampled transplant area(s) and calculating average density once the ramp tally is complete.
2022 data
Plot name total ramets scapes scape % notes
1. small yurt 129 27 20.9% harvested 2020
2. wood nettles E 138 45 32.6% harvested 2020
3. S. slope W 165 40 24.2% harvested 2020
4. wood nettles W 171 63 36.8% harvested 2020
5. First supertree 187 36 19.3% harvested 2020
6. Flat, slope base 196 29 14.8% harvested 2020
7. Basin slope 1 227 50 22.0% harvested recently, 2019 or 2020
8. Sap tank 239 38 15.9% not harvested
9. original V 255 30 11.8% light harvest 2020
10. Sap tank seed 398 18 4.5% planted 2012, never harvested
11. sledding hill 448 30 6.7% planted 2008, never harvested
dispersed transplants 1876 715 38.1% transplanted in 2020 from plots 1-6 and 9
Discussion
You can see that the relationship between plot density and scape production is quite strong. Plants growing at lower density are more likely to produce scapes. Since bulb division is largely dependent on scape production, we can see that both seed and vegetative reproduction will be increased at lower densities. This bears directly on the conclusion that after harvest (which lowers the density of ramets), reproduction rates increase, just as ecological theory predicts, and counter to the predictions modeled by Nault and Gagnon (1993).
Ramp Allium tricocum Demographics at High Densities in Fully Stocked Stands, and the Relevant Effects of Leaf Cutting.
This is a long-term study designed to answer two questions: 1) What happens to ramp populations over the long term in dense, fully-stocked populations that are not manipulated?
2) How does the harvesting of leaves impact the demographics of ramps in dense populations?
Well established theories of population ecology, as well as plain common sense and experience, indicate that a population in a given environment will reach a certain density of individuals or biomass per unit of area, after which this density should stop increasing due to limits imposed by available vital resources. We chose a long-established, dense, natural population of ramps from which we have never harvested to examine the long-term demographic trends in such a fully-stocked stand. In 2018, after leaf senescence, we established 2 adjacent meter-by-meter quadrats by driving spikes into the corners and running a cord around them to form quadrat boundaries. These quadrats are on a moderate west-facing slope, just downhill from a sugar maple Acer saccharum trunk. The north quadrat will not be manipulated, and will serve as a control from which to understand the demographics in a densely-stocked stand. By comparing this to lower-density or manipulated populations, we can examine density-dependent or manipulation-dependent differences in mortality, growth, and reproduction. The south quadrat will be subject to removal of one leaf yearly from each plant that has two or more leaves. This leaf removal will occur on the day of data collection, immediately after leaf number and size is recorded. We aim to do this after the leaves have reached approximate full size, but before senescence has begun. We intend to record the number of scapes well after the day of leaf counting because optimal time for leaf cutting (culinarily) is before the scape buds are visible. We will do our best to record data multiple times to avoid missing scapes or seeds, but timing has been problematic and some data is missing. In autumn we intend to record the number of viable seeds produced.
Data collection began in 2019 and will continue indefinitely.
2019 (leaf data 6-4, scape data 9-18)
control plot (N): 269 ramets, 1 scape
leaf-cutting plot (S): 279 ramets, 2 scapes
2020 (leaf data 5-9; scape data 5-20, seed data 9-23)
control plot (N): 244 ramets, 92 scapes, 16 umbels with 41 viable seeds
leaf-cutting plot (S): 274 ramets, 106 scapes, 24 umbels with 108 viable seeds
2021 (leaf data 5-5, scape data 6-27. Seed data not recorded)
control plot (N): 298 ramets, 22 scapes
Leaf-cutting plot (S): 264 ramets, 17 scapes
2022 (leaf data 5-19, scape data 6-1, Seed data 10-6)
C/N: 358 ramets, 8 scapes, no viable seeds
LC/S: 307 ramets, 3 scapes, no viable seeds
2023 (leaf data 5-20. No scape or seed data collected. Because root cellar)
C/N: 294 ramets
LC/S: 315 ramets
Smaller data sets and observations pertaining to specific ecological, demographic, and harvest questions regarding ramps Allium tricoccum.
Research questions and future study ideas.
This page is a work in progress. Please be patient. I have put thousands of hours into this, and thousands of dollars of my own money. Nobody is paying me for this. I have received no grants or other financial support from anyone. I have a busy life and will be updating and expanding the data here as time allows. Eventually I will probably set up an entire website for this research, after which this page will redirect readers there. Please email me with suggestions, concerns, ideas, etc.
Our research is primarily set in the Town of Hill, Price County, Wisconsin, at 45.5 N, 90.2 W, and approximately 550 meters of elevation, at our active, small, commercial maple sugaring operation. The terrain is terminal moraine with steep slopes in all aspects. The soil is glacial till with a moderate amount of clay and many large stones. The forest is dominated by sugar maple Acer saccharum (about 60% and increasing with management), white ash Fraxinus americana, basswood Tilia americana, American elm Ulmus americana, and yellow birch Betula allegheniansis, in respective order of dominance. Other trees present in small amounts include black cherry Prunus serotina, pin cherry Prunus pensylvanica, northern red oak Quercus rubra, red maple Acer rubrum, white bich Betula papyrifera, hophornbeam Ostrya virginiana, black ash Fraxinus nigra, eastern hemlock Tsuga canadensis, butternut Juglans cinerea, and slippery elm Ulmus rubra. The site was northern hemlock-hardwood forest with scattered white pine at the time of European settlement. The hemlock was removed in the early 1900s. The residual and second-growth hardwood stand was logged heavily about 1965, and most canopy trees in 2020 are 55-90 years old. This site is lucky to have adjacent old-growth stands for comparison, with most canopy trees in 240-315 years of age (as determined by growth ring counts on fallen trees). In these old-growth stands hemlock is the most dominant tree, and under the hemlocks no ramps or spring ephemerals are present. In some nearby hardwood stands with a dense ground layer of Canada yew Taxus canadensis the spring ephemeral herbs are also absent. The study site contains several small, robust patches of ramps and other spring ephemerals, such as Virginia waterleaf Hydrophyllum virginianum, Common spring beauty Claytonia virginica, and cut-leaf toothwort Cardamine concatenata, but is low in diversity and poorly stocked with typical herbs of a mesic northern hardwood forest. The few colonies of mesic forest herbs tend to be localized but large and nearly monocultural--not highly integrated with other species. This suggests recent and active colonization by these species, and means that most of the forest is ideal habitat for ramps, with very little competition, presumably due to the past dominance of hemlock and presenve of Canada yew precluding the spring ephemerals.
Basic Biology of Ramps, AKA Wild Leeks, (Allium tricoccum and Allium burdickii)
The taxonomy of ramps in eastern North America is contentious or complicated. For many years a single species, Allium tricoccum, was recognized. In 1979 this was split into two species, with the less common segregate being designated Allium burdickii. The latter species, sometimes called "white ramp," "green ramp," "small ramp," or "narrow-leaf ramp," is differentiated by being smaller, having narrower leaves, having a green (as opposed to reddish) leaf sheath, blooming about a month earlier, with scapes emerging well before leaf senescence. The umbels of A. burdickii are smaller and contain fewer flowers and seeds. The onion/garlic scent of its leaves is also weaker than that of the typical "red ramp" or "large ramp", ie, A. tricoccum. The common name "ramps" is a corruption of the English name "ramson" applied to a very similar and closely related mesic forest species of Europe.
In many locations the two North American ramp species are clearly differentiated. However, there are also extensive populations in which the characteristics that supposedly separate them are mixed. Some of these populations are hard to assign to either of the described species. Some researchers are currently looking at whether or not there are more than two species of ramps. This is not likely to be resolved by genetic work, however, because the definition and boundaries of species are probably the most problematic question in all of biology, and conclusions reside in the confluence of philosophy and fact. My personal working hypothesis is that we have two species of ramps that differentiated through past separation, but which have now come back into contact. Their processes of genetic isolation appear to be incomplete, perhaps more so in the northern part of their distributions, and currently it seems that they are re-converging, and some populations appear to be of hybrid origin. So, whether we consider them one species or several, suffice it to say that ramps are highly variable in size, phenology, and leaf characteristics, and their taxonomy is not settled. (And evolutionary theory implies that, in many cases, taxonomy should not be settled.)
Ramps are deciduous forest spring ephemerals; they emerge in early spring to take advantage of the brief growing period during which the canopy trees are devoid of leaves, and ample light hits the forest floor. Soon after the leaves appear on the overhead deciduous canopy, the ramp leaves senesce (die back) for the season. This makes for a typical growing season of only 4-9 weeks. However, scape production, flowering, seed production, and bulb division occur after the leaves have senesced, all relying on energy stored in the bulb. Flowering typically occurs in early to mid summer, and seeds ripen in late summer, although they may persist well into the fall, and even into the following summer. Seeds are dispersed by gravity, water, rodents, and some birds. The seeds do not germinate in their first spring; they germiante the second spring, about 19 months after maturing. Seedlings require at least seven years to reach sexual maturity (six years of growth, the first year being dormant). Bulb division occurs mostly in the late summer after the production of a scape; occasionally it occurs without scape production. Scape production rarely occurs unless a certain biomass is attained. After scape production, bulbs shrink or divide; they may flower in consecutive years, but more often store energy through one or more growing seasons before flowering again.
Ramps are found on many soil types of moderate to high fertility: clay, clay loam, loam, alluvial deposits of sand and silt mixtures, ground moraine, and glacial till. They are absent from very poor or thin soils, or those with very low capacity to retain moisture. They need neutral to slightly alkaline soils, and seem to thrive best where calcium levels are high. The classic tree associate is sugar maple, but ramps also thrive under basswood, red maple, siver maple, elms, yellowbud hickory, ash, yellow birch, buckeye, hornbeams (Carpinus and Ostrya), mulberry, butternut, balsam poplar, and many other species. Allium burdickii appears to thrive better in heavy clay soils. Allium tricoccum seems to grow largest on soils derived from limestone, or Appalachian slopes.
Definitions:
Ramet: This refers to a separate, independent biological unit of an organism that can reproduce asexually. Technically, each bulb in a clump of ramps may be genetically identical, though physically separate. The separate bulbs could be thought of as one organism broken into pieces, or ramets. Since there is no way to tell if a mature pre-existing bulb was produced by seed or bulb division, we call each bulb and its associated structures a ramet.
Study 1:
Population Growth and Demographics of Ramps Allium tricoccum at Low Densities: Analysis of Growth from Founder Bulbs Transplanted in 2007.
In this study we transplanted 85 ramp bulbs (each of which we call a founder bulb) to an area of suitable northern hardwood forest from which ramps were absent, so that we could monitor the population increase without the confounding variable of pre-existing plants. On May [ ], as the ramp leaves had reached full size, but before they had begun to senesce (die back) for the season, the bulbs were carefully dug out of one colony and immediately transplanted 30-60 meters away in two rows running up a rather steep south-facing hillside. The founder bulbs were placed approximately [ ] cm apart (although the distance was often not uniform due to obstructions such as logs and rocks). A small shovel was inserted about 6 inches (15 cm) into the soil and pulled back to create an opening into which the bulb was inserted about half way, allowing the roots to extend to the bottom of the hole. The soil was replaced around the bulb and roots, patted down, and less than half a liter of water was administered. This operation took less than one minute per transplant; after this, no care was given to the transplants. In 2012 the top broke off a white ash tree and landed in the west row—this top had a severe negative impact on five of the colonies but was allowed to decay in place rather than being removed. Although this study took place in an active commercial sugarbush, no thinning or other cutting of trees was done in the study area from 2007 to 2022.
In 2017 we assessed survival and reproduction. For each colony (defined here as the plants surrounding a founder bulb) we recorded each ramet (bulb), its number of leaves, and the width of each leaf at the widest point (to the nearest mm). We assigned each ramet as being derived from direct vegetative reproduction (bulb division), seed reproduction, or bulb division of a seedling. These categories were marked in our data as V, S, and VS respectively. The founder bulb was classified as V. (Although in some cases it was not possible to tell if the original founder bulb had died and was replaced by one of its daughter bulbs.) We also recorded the widest visible spread of seedlings from its apparent founder bulb.
There were a few uncertainties in our data collection. One is that it is impossible to know the seed source for any particular seedling—rain or gravity could have moved seeds between founder bulbs. This made it impossible to determine the greatest downhill spread of seedlings—an important weakness, because gravity should effect a greater spread downhill than in other directions. Thus, seed spread from founder bulbs was likely significantly underestimated. However, such movement within the plot would have had no effect on the totals for the study plot as a whole, and thus no impact on the calculated overall reproduction rate or demographics.
A second possibility is that seeds could have been transported into or out of the study plot by mammals or birds, as we know that one or more species consumes and transports the seeds (See Observations). However, since the density of ramps inside the study plot is greater than the overall average density of ramps on the surrounding landscape, both on an intermediate scale (5-50 meters, as is likely to be relevant for rodents), and on a broader scale (100-1000 meters, as is likely to be relevant for birds), the probability is that more seeds would have been transported out of the study area than into it. Thus there is reason to suspect that the observed reproduction may be slightly underestimated by outgoing seeds, and no reason to believe that it is inflated by incoming seeds.
Finally, a seedling that germinates close to the founder bulb can be mistaken for a daughter bulb. First-year daughter bulbs are much larger on average than seedlings at first emergence, and often have two leaves (as opposed to the single leaf always produced by new seedlings). Even when daughter bulbs have one leaf only, these are usually proportionately much broader than single seedling leaves. However, after 3-5 years these differences between daughter bulbs and seedlings are no longer evident. Since we did not collected data until 10 years after transplanting, it is probable that a small portion of the ramets designated as daughter bulbs were actually seedlings. Because of this uncertainty, we chose to designate all ramets with any bulb portion located within 2 cm of the founder bulb or founder clump (meaning the clump of ramets that arose vegetatively from the founder) as vegetatively reproduced. This means that, almost certainly, our data slightly inflates the percentage of reproduction allocated to direct vegetative reproduction, and underestimates the percentage allocated to seed.
Demographic data was collected again in 2018, 2020, and 2022. As the colonies expanded, the boundaries of the seedling zones around each founder became progressively harder to determine, so in the 2022 data collection, we tallied the lump sum of the ramets in each row, but no longer assigned ramets to any particular founder bulb or colony.
From our original 85 bulbs (ramets), the following totals were regenerated:
as of 2017:
2018: 1,324 ramets
2020: 1,937 ramets
2022: 2,851 ramets
Please allow time for further analysis. I have hundreds of pages of data for this study alone. I will break down the population by size class (leaf number), and also total leaf width (TLW), which is used as a proxy measure for biomass without having to kill the plants to weigh them. Since the original founder bulbs were all mature and had at least two leaves, while the later ramet tallies include a significant percentage of smaller seedlings, the percent increase of the TLW index should be somewhat lower than the figures for ramet number increase.
Another very important question that our data will address is the percentage of reproduction resulting from seed versus bulb division. I will post this as soon as I get the relevant totals.
Discussion
In 15 years, this is an increase of 33.54 times the original count, or a 3,354% increase. Please compare this to the increase rates modeled in Nault and Gagnon (1993). Note that Nault and Gagnon did not observe the population declines, they just predicted them, using a model designed to predict them, which could reasonably have no other outcome. Please note that these increases are based on very low initial population densities. Responsible harvesters would not reduce the population to the levels at which our study began, so the rate of increase should be lower in a harvested natural stand of similar quality.
Study 2:
Transplant Success at Different growth Phases, and the Effects of Competition from Pennsylvania Sedge Carex pennsylvanica on the Survival and Reproduction of Transplanted Ramps Allium tricoccum.
This study was designed with two purposes. The first was to test the relative success of transplanting ramp bulbs at 3 different growth stages: early emergence (leaves less than 3 inches long; this stage usually occurs at the study site for a brief period in April or early May), peak growth (leaves at full size, but before senescence begins; usually mid May), and summer dormancy (after the leaves have completely senesced). I established 3 plots, each divided into 3 rows of 15 bulbs each, for a total of 45 bulbs per plot. A few large obstructions, such as branches or rocks, were removed from each plot. In each plot, one row of bulbs was planted in each of the three growth stages, making for a total of 45 bulbs transplanted in each growth stage. The growth stage planted in each row was assigned randomly. Transplants for all 3 growth phases were pulled from the same nearby wild colony and sorted according to this protocol: Any daughter bulbs were separated, after which bulbs were selected for transplanting only if they were undamaged and had at least 6 roots 2 cm or more in length. After a large pool of qualifying bulbs was amassed in a tub, they were selected blindly, one at a time, by grabbing the first bulb or ramet to contact my fingers, and then shaking the tub between selections. Selected bulbs were placed in 3 separate piles of 15, after which a random numbers table was used to determine which pile was planted in which plot, as well as the order of planting the plots. The selection and transplanting for each growth phase was completed in one day, as rapidly as possible. A hole about 6 inches (15 cm) deep was opened with a trowel, a bulb was inserted with the lower terminus about 3 inches (8 cm) below the surface so the roots could rest in a natural position, and the soil was immediately pressed back into place over the bulb. This took about 20 seconds per bulb. No further care took place at any time.
While transplanting success or failure was apparent after two growing seasons, the study was designed to simultaneously test the long-term effects of competition from Pennsylvania sedge Carex pensylvanica. In two of the plots Pennsylvania sedge was absent or in very low density; the third plot was characterized by a dense covering of Pennsylvania sedge over the entire area. While transplanting success is important to recovery, management, and ecoculture practices involving ramps, the impacts of interspecific competition are also extremely important for such projects, and are even more important to understanding the ecology of this species in spontaneous wild populations.
Study 3:
Scape production rates among populations of ramps Allium tricoccum at high and low densities.
This study was designed to understand how the density of ramets in a population affects the production of scapes (flowering stems). Specifically, we are interested in whether or not the harvest of bulbs results in an increase in the scape production rate (and thus, reproduction and recovery) of remaining plants, as is predicted by the basic tenets of population ecology (Tilman, 1987; ); and contrary to the predictions of Nault and Gagnon (1993), who suggested that residual populations after harvest would show no increase in reproduction, and indeed might show increased mortality. The data was collected in late spring of 2022 (June 1), two growing seasons after bulbs had been harvested from some of the plots for transplanting. Unlike some of the projects, long-term monitoring for multiple years is not necessary to answer this question. However, additional sampling, as time permits, will strengthen the conclusions, and such data can be added at any future date, through replication of the sampling protocols we used, or alternative protocols which might have advantages.
We sampled in two separate ways. First, we placed quadrats of one meter squared, built from one-inch PVC pipe, into colonies of ramps, so that no portion of the quadrat was more than 30 cm from an established bulb. No quadrat placement overlapped any other previous placement. We sampled all of the colonies sufficiently large for quadrat placement within a designated section of our sugarbush; one colony was large enough to allow two quadrats. In total, 11 quadrats were sampled. A ramet was included in the quadrat if the upper terminus of its bulb was located within the quadrat; leaves leaning in from outside ramets were not counted. Bulbs directly under the quadrat border were included or excluded based on the natural location of their leaves.
Two of the sampled quadrats were in colonies established by seeding a decade or more prior to sampling, from which no ramets have been harvested. The other 9 quadrats were placed within spontaneously occurring wild colonies, 8 of which had been subject to some level of harvesting 2 years prior to sampling.
We counted each ramet in the quadrat, recording the number of leaves, and if a scape or scape bud was present. At this time some scapes were evident as buds in the juncture of the leaves, while other scapes had extended up to 14 centimeters. Although we checked systematically for scape buds, it is likely that we missed a few. However, past data collection indicates that nearly all scapes are evident at this stage of growth, and there is no reason to suspect a systematic bias in scape appearance between low and high density populations, so these few missed scape buds should have no material effect on the accuracy of the data. We did not record leaf width because 1) Some of the leaves were beginning to senesce, making accurate width measurement impossible, and 2) Although leaf width would add some value to the data, it was immaterial to the basic analysis we were aiming for.
Our sampling was for scapes only. We have consistently observed a very strong positive correlation between scape production and vegetative reproduction, and this relationship is borne out by the data of other researchers in all cases where it has been examined (Nault and Gagnon, 1993). We assume that there is also a positive correlation between scape production, flower production, and seed production. However, it is probable that this relationship is not linear, and our observations strongly suggest this. Scapes in lower-density populations appear to be larger on average, and to produce more flowers and seeds. Furthermore, due to seed predation, and the tendency of larger populations to attract vertebrate seed predators, or to harbor populations of insect seed predators, there also appears to be a complex relationship between scape density and successful production of viable seeds (see notes on study 2). Although the current study provides valuable insight into density-dependent reproduction of ramps, we intend to design and conduct research in the future to better understand potential differences in seed production per scape.
There is an important practical reason that we sampled for scapes only rather than flowers or seeds. Allium tricoccum is one of the few herbaceous species in our region that flowers when the leafy parts are dormant and thus not available for examination. If we were to count actual flowers or seeds, there would be no leaves available at the time of data collection, so we would not be able to effectively determine the density of non-flowering ramets in a plot, nor the leaf number, or leaf width, of these ramets. Our protocol allowed us to collect all of the data at once.
The obvious solution to this dilemma is to leave physical quadrats in place through the entire growing season, and collect the data at multiple times. This would allow us to see not only what percentage of ramets produce scapes, but also how many flowers and seeds were produced per scape. We plan to do this in the future, when we find the time to do it, perhaps at different study sites closer to home.
To the above 11 quadrats we add the 2022 data from two long-term quadrats in which we are monitoring ramet count, leaf count per ramet, and leaf width, as well as scape production (see study 4). These quadrats have seen no bulb harvesting; one of them has had leaves harvested.
The second means of data collection involved tracking the transplanted bulbs removed in 2020 from some of the colonies sampled by quadrat. These were transplanted to another section of the sugarbush with appropriate tree cover and soil, but totally devoid of ramps (and mostly devoid of spring ephemerals). They were transplanted on a rough grid, 50-70 cm apart, by opening a small hole about 15 cm into the soil with a trowel, holding a bulb with the bottom about 8 cm deep in the hole, and replacing the soil. This procedure took 15-20 seconds per bulb. No care was administered after transplanting.
We sampled this area by going down the rows of transplants and recording the number of ramets, their leaf counts, and the presence of scapes. Because of landscape idiosyncrasies that resulted in disorganization of the grid, we certainly missed many of the transplants, but there was no systematic bias that would threaten the reliability of the data. We did not calculate a density per meter squared for this area, but rather considered it “extremely low density, functionally equivalent to no intraspecific competition.” An average density per square meter in the transplant area would probably come to 4-7 ramets. However, in the future we will collect stronger and more readily comparable density data by marking the boundaries of the sampled transplant area(s) and calculating average density once the ramp tally is complete.
2022 data
Plot name total ramets scapes scape % notes
1. small yurt 129 27 20.9% harvested 2020
2. wood nettles E 138 45 32.6% harvested 2020
3. S. slope W 165 40 24.2% harvested 2020
4. wood nettles W 171 63 36.8% harvested 2020
5. First supertree 187 36 19.3% harvested 2020
6. Flat, slope base 196 29 14.8% harvested 2020
7. Basin slope 1 227 50 22.0% harvested recently, 2019 or 2020
8. Sap tank 239 38 15.9% not harvested
9. original V 255 30 11.8% light harvest 2020
10. Sap tank seed 398 18 4.5% planted 2012, never harvested
11. sledding hill 448 30 6.7% planted 2008, never harvested
dispersed transplants 1876 715 38.1% transplanted in 2020 from plots 1-6 and 9
Discussion
You can see that the relationship between plot density and scape production is quite strong. Plants growing at lower density are more likely to produce scapes. Since bulb division is largely dependent on scape production, we can see that both seed and vegetative reproduction will be increased at lower densities. This bears directly on the conclusion that after harvest (which lowers the density of ramets), reproduction rates increase, just as ecological theory predicts, and counter to the predictions modeled by Nault and Gagnon (1993).
Ramp Allium tricocum Demographics at High Densities in Fully Stocked Stands, and the Relevant Effects of Leaf Cutting.
This is a long-term study designed to answer two questions: 1) What happens to ramp populations over the long term in dense, fully-stocked populations that are not manipulated?
2) How does the harvesting of leaves impact the demographics of ramps in dense populations?
Well established theories of population ecology, as well as plain common sense and experience, indicate that a population in a given environment will reach a certain density of individuals or biomass per unit of area, after which this density should stop increasing due to limits imposed by available vital resources. We chose a long-established, dense, natural population of ramps from which we have never harvested to examine the long-term demographic trends in such a fully-stocked stand. In 2018, after leaf senescence, we established 2 adjacent meter-by-meter quadrats by driving spikes into the corners and running a cord around them to form quadrat boundaries. These quadrats are on a moderate west-facing slope, just downhill from a sugar maple Acer saccharum trunk. The north quadrat will not be manipulated, and will serve as a control from which to understand the demographics in a densely-stocked stand. By comparing this to lower-density or manipulated populations, we can examine density-dependent or manipulation-dependent differences in mortality, growth, and reproduction. The south quadrat will be subject to removal of one leaf yearly from each plant that has two or more leaves. This leaf removal will occur on the day of data collection, immediately after leaf number and size is recorded. We aim to do this after the leaves have reached approximate full size, but before senescence has begun. We intend to record the number of scapes well after the day of leaf counting because optimal time for leaf cutting (culinarily) is before the scape buds are visible. We will do our best to record data multiple times to avoid missing scapes or seeds, but timing has been problematic and some data is missing. In autumn we intend to record the number of viable seeds produced.
Data collection began in 2019 and will continue indefinitely.
2019 (leaf data 6-4, scape data 9-18)
control plot (N): 269 ramets, 1 scape
leaf-cutting plot (S): 279 ramets, 2 scapes
2020 (leaf data 5-9; scape data 5-20, seed data 9-23)
control plot (N): 244 ramets, 92 scapes, 16 umbels with 41 viable seeds
leaf-cutting plot (S): 274 ramets, 106 scapes, 24 umbels with 108 viable seeds
2021 (leaf data 5-5, scape data 6-27. Seed data not recorded)
control plot (N): 298 ramets, 22 scapes
Leaf-cutting plot (S): 264 ramets, 17 scapes
2022 (leaf data 5-19, scape data 6-1, Seed data 10-6)
C/N: 358 ramets, 8 scapes, no viable seeds
LC/S: 307 ramets, 3 scapes, no viable seeds
2023 (leaf data 5-20. No scape or seed data collected. Because root cellar)
C/N: 294 ramets
LC/S: 315 ramets
Smaller data sets and observations pertaining to specific ecological, demographic, and harvest questions regarding ramps Allium tricoccum.
Research questions and future study ideas.