A review of the ecology and population biology of Goldenseal, and protocols for monitoring its populations

Final report to the Office of Scientific Authority of the US Fish and Wildlife Service

Dr. Daniel Gagnon
Groupe de recherche en écologie forestière Université du Québec á Montréal

22 March 1999

Table of Contents

Executive Summary

1. Introduction

2. A literature review of goldenseal ecology and population biology

3. What is the likely impact of harvesting on goldenseal populations?

4. Biological and socio-economic reasons why the harvest of wild goldenseal populations  is probably not sustainable

5. Recommendations for goldenseal monitoring and conservation
    5.1. Inventory and monitoring
    5.2. Conservation
    5.3. Agriculture
    5.4. Research

6. Monitoring wild goldenseal populations
    6.1. Two levels of monitoring intensity for two different purposes
    6.2. Monitoring for general population trends
    6.3. Monitoring for population dynamics information
        6.3.1. Particular difficulties of goldenseal demographic monitoring
        6.3.2. Study population selection and time required
        6.3.3. Study population site and vegetation description
        6.3.4. Demographic data sampling
  Locating individual plants
  Data collected for individual plants
  Sampling date and frequency, and study duration
         6.3.5. Field and garden seed germination trials
    6.4. Overview of demographic analysis methods


Appendix 1: “Site and ecological data” field form

Appendix 2: “Vegetation data” field form

Appendix 3: “Goldenseal data” field form
Figure 1.  Diagram of goldenseal size/state classes and field measurements for demographicanalysis

Executive summary

Goldenseal (Hydrastis canadensis L.) has just recently been listed in CITES Appendix II. The species is designated as Threatened in Canada (COSEWIC), with priority 1 ranking (the highest) for protection, where it occurs only in the extreme southern parts of Ontario.  In the United States, a national program to regulate export and monitor wild populations will be implemented by the US CITES Authority, the US Fish and Wildlife Service.

The current knowledge about the population biology of goldenseal is briefly reviewed in this report, based mostly on information from literature on its cultivation.  There are no population dynamics data available for goldenseal, from anywhere within its range.  There is therefore a serious lack of scientific data on the population ecology of goldenseal in the wild.  The impact of harvesting on goldenseal populations can nevertheless be inferred to be detrimental, based on comparisons with other clonal forest perennial herbs.

Although goldenseal is harvested in the wild in the US, there is no systematic monitoring of wild populations.  There are biological reasons, which are common to all slow growing forest understory herbs, like goldenseal and ginseng, as well as present day socio-economic reasons, that raise some doubts about the sustainability of harvesting wild goldenseal roots.  These built-in impediments to sustainability are briefly reviewed in this report.  There is a potential for developing a lucrative agricultural industry based on this species, which may never come about if natural populations, the gene pool of the species, were to disappear.  Cultivation of goldenseal is a good option for relieving harvesting pressure on wild populations, particularly if a moratorium on the exportation of wild goldenseal was to be put in place.

In spite of widespread indications that the reality of the field situation may be very poor, accurate and detailed information is lacking in order to provide conclusive evidence of sustainability or lack of sustainability for wild goldenseal harvesting.  Recommendations for action are provided in this report along the lines of monitoring, conservation, agriculture and research, in order to improve the situation of goldenseal in the US.

The major part of this report describes a two-tiered goldenseal population monitoring program, and relevant field protocols, which should provide most of the answers needed to manage the goldenseal resource for conservation, and perhaps continued harvesting. The monitoring program will yield population dynamics information from the high intensity monitoring of a smaller number of populations, and general population trends from the low intensity monitoring of a larger number of populations.  Hopefully, this monitoring program will be implemented in partnership with the states and the US Fish and Wildlife Service, as well as other partners.


Goldenseal (Hydrastis canadensis L.) has just recently been listed in CITES Appendix II. The species is designated as Threatened in Canada (COSEWIC), with priority 1 ranking (the highest) for protection, where it occurs only in the extreme southern parts of Ontario.  In the United States, a national program to regulate export and monitor wild populations will be implemented by the US CITES Authority, the US Fish and Wildlife Service.

Goldenseal occurs in the US on rich and moist soils of deciduous forests, from southern New England west to southern Wisconsin, south to Arkansas and northern Georgia.  Its habitat is much like that of ginseng (Panax quinquefolius).  However, goldenseal was only ever abundant in the central portion of its range in the states of Indiana, Kentucky, Ohio and West Virginia (Catling and Small 1994).  Collecting for the drug trade began in the mid 1800s, and has certainly played a role in drastically reducing wild populations, although habitat loss was also a factor in its decline.  Traditionally, goldenseal was mostly harvested from the mountains of Kentucky and West Virginia by people whose economy included the harvest of several wild crops (ginseng, May-apple, etc.).

In the late 1800s approximately 65,000 kg were collected annually (about 550 dry roots per kg), and destined then almost entirely to the North American market (Catling and Small 1994).  In this respect it was very different from wild harvested ginseng, whose market was essentially Asian.  If this pattern has persisted to the present (no particular value placed on wild roots by consumers), it is likely to facilitate efforts to base future market requirements for goldenseal more and more on agricultural production, rather than wild harvests.  Limited available information suggests that goldenseal may be easier to cultivate than ginseng, tolerating higher light levels and having fewer pests and diseases.  Even though the current supply of goldenseal still comes mostly from wild harvested roots, it has already been widely cultivated (Arkansas, Michigan, North Carolina, Oregon, Tennessee, Washington, Wisconsin).  Ginseng growers can use their shading installations to grow goldenseal, allowing for crop diversification.

Although the harvesting history of goldenseal seems to be adequately known (through quantities sold), it would seem that the same cannot be said for information about the population biology of the species.  Such knowledge is normally the common basis for the management of any harvested wild species (i.e. big game and sports fishing species).  There is no existing population dynamics data for goldenseal, from anywhere within its range.  Information is equally scant for other aspects of the ecology of this species (abundance of populations, size of populations, habitat characteristics).  There is also no systematic monitoring of any of its wild populations (USFWS/OSA 1997).

The purpose of this proposal was to search extensively the existing literature in order to find and summarize all of the available scientific information on goldenseal, particularly as it pertains to its population biology, but also to other aspects of its ecology.  This search has turned up nothing new on the ecology of goldenseal.  Using the limited available information, I wrote an overview of the population dynamics of the species, as well as a discussion of the likely impacts (and sustainability) of harvesting wild goldenseal, based on comparisons with other forest perennial herbs.  Recommendations are proposed for the monitoring and the conservation of goldenseal populations.  Detailed protocols for two levels of goldenseal population monitoring are included.  This report provides a scientific basis for initiating a monitoring program of goldenseal populations, and for the eventual clarification of its population dynamics and vulnerability to harvests.

2. A literature review of goldenseal ecology and population  biology

A literature search, covering from 1988 to 1999, found five references to goldenseal (four when searching with “goldenseal”, one when searching with “Hydrastis canadensis”), none of which were pertinent to the ecology or population biology of the species.  Similarly, the COSEWIC report for goldenseal in Canada (White 1991) and the proposal to include goldenseal in Appendix II of the CITES agreement (USFWS/OSA 1997) contain no published references to goldenseal ecology, except for only two papers.  The most recent paper deals with distribution, phenology and biomass of goldenseal in the understory of an Indiana oak-hickory forest (Eichenberger and Parker 1976).  Another paper, published in 1891, discusses the life history of goldenseal (Bowers 1891).  This represents the sum total of what has been published about the ecology of natural populations of goldenseal during the last 100 years.

With such a list, a literature review can be very short indeed.  To my knowledge, no study has been made of the population dynamics of natural goldenseal populations.  However, Dr. Jim McGraw from the University of West Virginia will be initiating a research program on this species in the wild starting in 1999 (J. Alvarez, pers. com.).  This research should produce sound information about the population biology of goldenseal.  All of the current useful information about goldenseal biology and ecology comes from agricultural literature, most notably the USDA farmers’ bulletin written by Van Fleet (1914, revised in 1949) and the more recent paper by Davis (1995).  This information is summarized below.

Hydrastis canadensis is a long-lived perennial herb of deciduous forest understories.  It occurs in the Eastern North American deciduous forest.  The habitat it occurs in, and its associated species (White 1991) are very similar to that of ginseng.  The underground parts of the plant consists of a horizontal, gnarled, structure of rhizomes, which bears the scars from the yearly abscission of the aerial stem.  The rhizome is covered in long fibrous roots.  As the plant grows, the rhizome will increase in size only up to a certain point, because it breaks up into other plants (or ramets, i.e. genetically identical and vegetatively produced individuals, which may or may not retain root or rhizome connections) through clonal growth (Van Fleet 1949).  Each annual section of the rhizome is a potential growing point for a new ramet.  Two buds are usually present at the base of an aerial stem, one of which will replace the stem in the following year.  The second bud appears to act as a reserve bud, to replace the first bud in case of damage (Van Fleet 1949), although Eichenberger and Parker (1976) report that many growing points had two aerial stems at their Indiana study site.  Vegetative propagation is also possible through budding on the long fibrous roots that emerge laterally from the rhizome (Van Fleet 1949).  This well developed vegetative propagation results in dense patches of goldenseal plants, of up to 100 stems in a single square meter (White 1991).  Vegetative propagation (clonal growth) will produce high genetic uniformity within most populations.

Depending on its size, a goldenseal plant will have from one to three leaves.  These leaves are palmate, broadly rounded and five to seven-lobed (more conspicuously five-parted).  The stem and petioles are pubescent; leaves are only slightly hairy.  Leaf emergence is from mid- to late April, and leaves are usually fully developed by mid-May.  Emerging seedlings have one small leaf (seed-leaf stage) (Van Fleet 1949).  In subsequent years they may have one or two of the characteristic palmate leaves, emerging from the same growing point.  Larger plants will have a stem bearing two leaves (a large lower leaf and a smaller upper leaf) and also often a third basal leaf emerging from the same growing point as the stem.  Above the smaller stem leaf is the flower stalk, bearing a single greenish white flower.  Flowering is in early spring, and lasts only a few days.  By early July, a fleshy fruit appears, raspberry-like in appearance, which will turn bright red and ripen during July and August.  Each individual berry in this aggregate contains a single shiny black or dark brown seed (2.5 mm in length).  Thus, each flower may produce 10 to 30 seeds (Van Fleet 1949).

Vegetatively produced plants will have one or two of the characteristic leaves, the size of which makes it impossible to confuse them with seedlings.  Observations in southern Ontario goldenseal populations (in 1989) revealed the presence of few flowers and fruit (White 1991).  Most populations observed had dense patches of plants with limited seed production, suggesting that vegetative propagation was the dominant means of population maintenance and expansion (White 1991). Van Fleet (1949) reports that in old (dense) clumps, as well as on young or weak plants, there are many stems that bear a single leaf and no flower.

All of the above information is basic knowledge that is very useful in designing a goldenseal population dynamics or monitoring program.  However, none of this information gives any indication of vital rates (vegetative division rate, seedling recruitment rates, mortality rates, individual growth rates) that are necessary to make population growth projections.  Without these data, neither can we estimate how much harvesting a goldenseal population is likely to be able to tolerate.  It is therefore imperative to initiate population dynamics studies (high intensity level monitoring, see section 6.3) in natural goldenseal populations.

3. What is the likely impact of harvesting on goldenseal  populations?

This question can be answered because of what is already known of goldenseal biology through its cultivation, and the type of habitat that it occupies.  Goldenseal, as ginseng (Charron and Gagnon 1991; Nantel et al. 1996) and ramps (Nault and Gagnon 1993), is a forest understory herb.  The forest understory is a relatively stable environment (small changes or variations from year to year) where the plants are adapted to growing in the shade (except of course the spring flowering plants), and face competition for soil nutrients and water from other understory plants (herbs, shrubs and tree seedlings) as well as from the overstory trees.  These plants may preyed upon by herbivores (insects, mammals) which periodically eat their leaves (partially or entirely) and fruit/seeds.  Some plants are even dug up by animals which consume the roots.  However, in the case of goldenseal, there appears to be little herbivory.  This is undoubtedly because of the powerful alkaloids found in all parts of the plant, which would act as strong deterrents.  This postulated herbivore defense of goldenseal is achieved through the costly synthesis of secondary compounds.  These come at a metabolic cost to the plant, and are therefore expected to reduce the plant’s growth rate.

It stands to reason that in a forest understory environment (light stress, high competition), no plant would be able to produce large quantities of biomass.  The first requirement for building plant biomass is sunlight.  Deciduous forest understories are notoriously dark, allowing only 5 to 15 % of sunlight to reach the forest floor.  Conversely, goldenseal growers use 70 to 75 % artificial shading over their crop, which therefore receives 30 to 25 % of the sunlight.  This increase in available light (from 5-15 % to 25-30 %) boosts productivity.  Similarly, even though goldenseal occurs in soils that are nutrient rich (for forest soils) and moist (except during occasional extended periods without rain), the soils are fertilized in the agricultural situation and may be irrigated in periods of drought.  These soil factors again boost productivity.  Growers also control fungal pathogens.  Agriculture can produce a marketable goldenseal root in three or four years.  A goldenseal plant growing in the wild will require at least four to five times that amount of time to produce a root of equivalent size, and this is assuming optimal forest growing conditions.

Goldenseal is not the only slow growing forest perennial herb.  Other studies have reported population growth rates near equilibrium (1.00) for many perennial understory plants.  Examples are Panax quinquefolius (American ginseng) with a population growth rate varying between 0.88 and 1.17,Chamaelirium luteum (fairy wand), with a growth rate varying between 0.99 and 1.06, Arisaema triphyllum (jack-in-the-pulpit) with a growth rate varying between 0.85 and 1.32, Arisaema serratum (a Japanese jack-in-the-pulpit) with a growth rate of 0.99, and Allium tricoccum (wild leek or ramp) with a growth rate varying between 1.02 and 1.13 (Charron and Gagnon 1991; Meagher 1982; Bierzychudek 1982; Kinoshita 1987; Nault and Gagnon 1993).  Conversely, studies of herbs from non-forested habitats have reported extremely variable population growth rates: 0.58 - 1.81 in Pedicularis furbishiae; 0.28 - 2.61 in Dipsacus sylvestris (teasel); 0.77 - 1.65 in Eriophorum vaginatum; and 0.60 - 1.43 in Danthonia sericea (Menges 1990; Werner and Caswell 1977; Fetcher and Shaver 1983; Moloney 1988).  It would appear that this high variability is associated with open and disturbed habitats, where the potential for fast growth is present when the habitat has just been disturbed (much light, soil nutrients and few competitors).  We then see these plants, such as the field weed teasel, having huge population growth rates (i.e. 2.61).  However, after a few years, when exceptional initial conditions have changed (encroachment by other plant competitors), the growth rates fall much below 1.00 and the population is fast declining to extinction.

As mentioned before, goldenseal is a plant that grows in a relatively stable forest understory habitat.  It probably has a population growth rate that varies relatively little, above and below 1.00 (population stability or maintenance).  The surplus growth above 1.00, translatable in individual plants, would represents the harvestable proportion of the population.  Any harvest that removes more than this “surplus” number of plants basically eats into the “capital” of the population.  The “surplus” in some years may also be necessary to compensate for losses that occur in “bad” years (poor growing season).  There is no logical reason why goldenseal populations would behave differently than any other of the forest perennial herbs already studied.  Therefore, harvesting impacts are likely to be serious on any goldenseal population, and only sufficient recovery time (several years, perhaps a decade) between harvests can make it sustainable.  This lengthy recovery scenario is extremely unlikely to occur nowadays.

4. Biological and socio-economic reasons why the harvest of  wild goldenseal populations is probably not sustainable

Biologically, harvesting of goldenseal makes little sense.  The root of the plant is harvested, effectively killing it (not at all like harvesting wild fruit, such as blueberries, or even leaves, such as those of the fiddlehead fern, Matteucia struthiopteris).  A yearly harvest is evidently not sustainable, considering the slow growth of the plant.  Goldenseal appears to propagate mainly vegetatively, and some useful comparison may be made with ramps (Allium tricoccum) where this is also the case (Nault and Gagnon 1993).  Removal experiments (simulated harvesting) on ramp populations in the Great Smoky Mountains National Park have shown that recovery from 50 % harvesting of ramp ramets is not complete even after 10 years (J. Rock, unpublished manuscript).  Stochastic harvesting simulations have also revealed that ramps are highly sensitive to harvesting, just as equally as ginseng (Nantel et al. 1996).  Furthermore, the minimum viable population size of ramps was estimated at 1000 plants, whereas ginseng’s MVP size was estimated at 172 plants.

Why would a vegetatively propagating plant like ramps (and goldenseal) be just as vulnerable to harvesting as ginseng, which reproduces exclusively from seed?  Because vegetative propagation is a highly successful means of establishing new plants (clonal individuals), as opposed to the establishment of seedlings, which has a high level of failure.  Vegetatively propagated plants have rhizome reserves (or at least a root connection with a larger plant) which considerably increases survival (comparable to that of mature plants) and potential for growth, even under adverse conditions (ex. competition).  Therefore, when underground structures of plants that propagate vegetatively are harvested, a tremendous loss of future growth capital is destroyed.  Wild leek (ramp) populations declined and disappeared rapidly in Quebec under commercial harvest pressure, leading the province to officially declare wild leek a vulnerable species (under its Endangered and Vulnerable Species protection law) and prohibit all sales of the plant in 1995.

The 1997 proposal to CITES quotes observers as saying that entire populations of goldenseal are extirpated by diggers (USFWS/OSA 1997). Perhaps in the past, a harvester would follow a route known only to him between numerous populations, and harvest just enough to allow each population to recover.  Nowadays, such behavior is likely to be counter-productive if another (or several) less scrupulous harvester finds the populations and removes all plants.  There are no signs that goldenseal is becoming easier to find, more to the contrary (see several quotes in USFWS/OSA proposal, 1997).  This brings us to the change in socio-economic factors, which may have, at some point in the past, been appropriate to sustain a harvest of goldenseal.

The socio-economic fabric of areas where goldenseal is still harvested has changed dramatically from that of pioneer days, when goldenseal harvesting was one of many subsistence activities done over vast tracks of virgin forests by a small human population.  The story of the careful harvester (because he left behind still healthy populations that he intended to return to) who did large circuits every fall to several large populations, not even visiting all the ones he knew every year, is a definitely a thing of the past.  This idealized vision of the past harvester may also not be true.  Perhaps there simply was not enough people digging goldenseal in the past to deplete the resource.  Or maybe the income produced was not sufficient to support a family, and time was better devoted to other activities, such as hunting and agriculture.  The value of goldenseal roots was certainly not as high as it is today, until recently.

Harvesting goldenseal from the wild in the early 1800s was not an activity for the faint of hearth.  The risks were numerous.  Today, harvesters drive to a trail head and return at the end of the day to drive back to the comfort of their home.  However, nobody can earn a living harvesting goldenseal.  Populations are too few and far between, and they contain few plants.

The harvest of goldenseal is not really sustainable in a biological sense, unless harvested populations are not entirely decimated and are left to recover for numerous years.  We know that such a scenario is very unlikely.  Nowadays, most populations are harvested in their entirety when found.  There are nowadays more people who know of the high value of goldenseal and who are out there looking for it, even as a weekend activity.  Accessibility is certainly not a problem, with vehicles, and a large network of roads and trails.  Suitable habitat may also be decreasing because of recent increases in logging.  Habitat fragmentation may prevent goldenseal from recolonizing sites where it used to grow.

The whole socio-economic picture can be summarized by the statement that an increasing number of people are seeking an ever decreasing resource.  Field verification is necessary to support this assumption of diminishing population sizes and of populations disappearing altogether.  Perhaps when field investigators see monitored populations disappear one after the other, will a clear and scientifically valid verdict on the sustainability of goldenseal harvesting from the wild be reached.

What about limits and regulations?  These can be valuable only if they can be enforced.  Personal limits on how much goldenseal any individual can have in his possession at any one time can be enforced (with good record keeping), but they can also be circumvented by having relatives sell some roots.  On the other hand, all regulations and limits pertaining to harvesting populations are clearly impossible to enforce, and therefore useless.  It would be impossible to prove that a digger had harvested all of the plants in each population he visited.

5. Recommendations for goldenseal monitoring and conservation

The recommendations below are suggested to reduce negative impacts of harvesting on goldenseal populations, to increase knowledge about the population dynamics and actual field status of goldenseal in the US, and to enhance its conservation.

5.1. Inventory and monitoring

5.2. Conservation 5.3. Agriculture 5.4. Research 6. Monitoring wild goldenseal populations

6.1. Two levels of monitoring intensity for two different purposes

Monitoring can mean very different things to different people, or require very different things depending on what we would like to demonstrate or achieve with its results.  A good framework is provided by a recent paper by Menges and Gordon (1996), entitled “Three levels of monitoring intensity for rare plant species”.  The authors list four general goals to monitoring biological resources: (1) detecting significant changes in abundance; (2) understanding the reasons for these changes; (3) determining the effects of management practices (including harvesting) on population dynamics; and (4) suggesting key applied research questions (Menges and Gordon 1996).

Goldenseal is a very valuable biological resource.  It is one the most sought after plant in the entire US and its harvest from the wild brings in a sizable revenue.  It should therefore be important that its status in the wild be accurately assessed.  Are wild populations of goldenseal declining? In numbers of populations? In numbers of plants?  Or are they maintaining their numbers?  Is this picture the same everywhere in the US?  Nothing can replace actual field verifications for these types of data.  No wildlife biologist would accept making hunting harvest quotas without some field surveys to assess abundance, and without knowledge of the regional demographic characteristics of the population he or she has to manage.

Acquiring knowledge about abundance and about demographic characteristics requires two different levels of field monitoring, the second and third level described by Menges and Gordon (1996).  Their first level requires only the recording of the presence of a population, and reveals nothing about the state of each population, until it is no longer there.  The second level of monitoring requires a quantitative assessment of abundance.  In the case of goldenseal this could be a count, by leaf number/flower size-classes, of all the plants in all the populations monitored at this level (see section 6.2 below).  This level of monitoring allows the analysis of population trends.  It answers questions about the numbers of populations and plants, and after a number of years of monitoring, general or individual population trends can be detected (increase, decline, overall stability).  The third level of monitoring, which is the most intensive, requires the monitoring of marked individuals for several years, usually over 100 in each study population (see section 6.3 below).  This level of monitoring provides quantitative information about demographic parameters.  The data can be used for calculating population growth rates, for making population viability analyses, for modeling the effects of harvesting or management practices, etc.  Demographic mechanisms underlying population trends can then be identified.

A characteristic of goldenseal population monitoring, as is also the case for ginseng, is the level of confidentiality, or even secrecy, involved in the record keeping of accurate locations of the existing or monitored populations.  Populations identified on maps may disappear very rapidly if the information reaches unscrupulous people.  Because of this, private land owners and even public servants may be very reluctant to divulge the location of the populations they know.  Therefore, good security and very limited access to precise site information is essential for success.  Otherwise, the monitoring process may itself cause the disappearance of goldenseal populations.

6.2. Monitoring for general population trends

A maximum number of goldenseal populations should be monitored for basic population trends.  Assuming each state agrees to support the intensive monitoring of 10 goldenseal populations, at a level necessary for population dynamics information (see section 6.3 below), an additional 20 to 30 goldenseal populations should be monitored for general population trends.  Because the data of intensively monitored populations can be used as well for general population trends, each state would then have general population trend information for 30 to 40 populations.

As the population selection guidelines are identical for both types of monitored populations, these are given in the following section (6.3.2).  Similarly, site data and vegetation data for each monitored population are required, in order to be able to determine if the intensively monitored populations are representative of the majority of goldenseal populations in the state (see section 6.3.3, and field forms in appendices 1 and 2).

The only difference between the two types of monitoring, is that the less intensive population trend monitoring does not require the year to year precise identification of individual plants.  Initial sampling is thus much less time consuming, as no precise mapping is required for individual plants (a general mapping of patches of plants within the population is however useful for subsequent monitoring).  Similarly, when the populations are revisited, no time is required to verify the identity (number and position on map and/or micro-plot) of each individual plant.  All that is required, every year as the population is monitored, is recording the number of plants found in each state or size-class category.  The number of seedlings and larger one-leaved plants may vary quite a bit from year to year, depending on yearly environmental variations, but also because of the varying thoroughness of each investigator (especially if they change from year to year).  However, this lack of precision in recording plants in the smaller size-classes is not a serious problem, as all investigators are likely to record all goldenseal plants found that are larger than seedlings.  Goldenseal may also recruit very few new plants into its populations via seedlings, as vegetative propagation appears to dominate.  The year to year comparison of the state/size-class structure (a graph of how many plants are found in each state/size-class) of a population will reveal a large amount of information, such as population growth (increase in number of plants) and evidence of harvesting events (sudden disappearance of all two-leaved plants or larger).  Following a harvesting event, it will be very useful to continue monitoring the population to record its regrowth, or perhaps its eventual decline to extirpation.

Soon (within a few years), an extremely valuable baseline of information will have been acquired on how many populations have been harvested, how frequently, and how many have regained their numbers, or declined to smaller population size, or even declined to extinction.  A record should be kept of all known populations and their size (an inventory), and their continued existence should be verified every few years (i.e. all are revisited within three to five years).  This is equivalent to the level 1 monitoring of Menges and Gordon (1996).  It can send a coarse warning signal if suddenly a large percentage of known populations are no longer found.  Some of these populations may be monitored for general population size trend in the future, in replacement of previously monitored populations that have since disappeared.

On top of recording the number of plants per size- or state-class, general population monitoring would benefit from a recording of each plant’s fruit production, and even the number of seeds.  The same field form as for intensive monitoring can be used to record this data (appendix 3), the only difference being that each goldenseal plant is not identified precisely (saving considerable time).  The time required to complete this low intensity monitoring is approximately half of what is required for the high intensity monitoring (described in section 8.3).  The time difference is not greater simply because monitored populations are not expected to be located just next to each other, and that some car travel and hiking time will be required in between populations.  Therefore, it is expected that a team of two field investigators should be able to record all of the necessary data for each population in half a day.  There should be no difference in this estimate between the first year and subsequent years,  because site and vegetation data can be recorded in about half an hour (during the first visit).  A field team should be able to monitor 20 to 30 populations in 10 to 15 working days, with no allowances for inclement weather, as data forms are uncomplicated and fast to complete.  The optimal time of year for this type of monitoring is identical to that of the more intensive monitoring, but because it not so critical to accurately record fruit and seed production (the numbers of plants in each state/size-class are the most important data), this monitoring could be carried out before and after the optimal period for demographic monitoring.  In fact, population trend monitoring could start in the first week of July and serve to decide when the more intensive demographic monitoring could start (by observing daily the state of fruit ripening).  When close to half the fruit-bearing plants have fruit that has turned red, the intensively monitored populations could be all sampled (10 in a two to three weeks period, by one sampling team of two, or in one to one and a half weeks by two sampling teams).  Afterwards, the low intensity monitoring could be resumed.

In all, 30 low intensity monitoring populations and 10 high intensity monitoring populations could be sampled in 5 weeks (from the first week of July to the end of the first week of August) by a two person sampling team, during the first year of the monitoring program.  Two sampling teams may be required in the first year if distances are large, and if populations are not already precisely located.  Ideally, this sample size should be acquired in each state.  However, time constraints, budget constraints and lack of adequately trained personnel may force some states to initially aim for smaller objectives (or spread them over two years).  A sufficient number of populations may also not be known to state officials.  This monitoring program should at least extend to those states where goldenseal harvesting is permitted, but the participation of states where goldenseal is not harvested or rare would also be very useful and should be solicited and encouraged.

6.3. Monitoring for population dynamics information

6.3.1. Particular difficulties of goldenseal demographic monitoring

Goldenseal is a clonally propagated species, and this causes several difficulties in following the fate of individual plants over a number of years.  First of all, each plant is probably not an “individual” in the genetic sense (or “genet”), but is more likely to be a “ramet”, which is the product of vegetative division.  Therefore, several nearby plants may be genetically identical, even though root or rhizome connections have decayed and they are separate plants.  This kind of propagation (budding off of rhizome and roots) leads to dense patches of plants, often 100 per square meter (White 1991).  This makes it difficult to identify individual ramets from year to year, unless careful mapping is done within micro-plots (usually 1m x 1m).

Vegetative growth also means that new ramets will appear.  Those new vegetative recruits will need to be mapped every year.  It is also important that their “mother plant” be identified.  How many and how often vegetative offshoots are produced by plants of different sizes is important demographic data.  In order to confirm daughter and mother plant connections, leaf litter and superficial humus may need to be removed carefully in order to reveal roots or rhizomes, and replaced afterwards.  Demographic analysis of clonal plants has been done before, but it requires more field effort and data are more complicated to analyze.  An example is ramps (wild leeks, Allium tricoccum), where the onion-like bulbs divide to produce one or two new vegetative recruits (Nault and Gagnon 1993).  An advantage of ramps is that the new plants appear right beside the large plant that was present the previous year.  With goldenseal, vegetative recruits can appear on any part of the rhizome or on the roots, at various distances from the mother plant, which makes it harder to identify the mother plant.  However, similar efforts were required for the study of a clonal wild sunflower (Helianthus divaricatus) which produced two offshoots every year that emerged 10 to 20 cm away from the mother plant (Nantel and Gagnon 1999).  Some amount of surface soil removal was necessary every year to confirm the origin of the vegetative offshoots.

6.3.2. Study population selection and time required

The study of the population dynamics of goldenseal should be based on the study of as many populations, of over 100 individuals each, as possible.  The limitation will be the resources available to do the annual monitoring.  The first year effort will undoubtedly be greater because the precise mapping of all individuals is required.  More time will be required also if a search is necessary to locate study populations.  However, the staff of national parks, state parks, national and state forests, as well as owners of private land or protected areas may know of some populations that meet the criteria for this level of monitoring.  The field data collecting itself should be done by two person teams (one assessing each plant, and one taking notes and mapping).  A team is expected to sample at least a population per field day, with allowances for driving and hiking to the site.  If populations are close, and easily accessible, two populations sampled per day is feasible.

Assuming a state has the responsibility of monitoring 10 populations for population dynamics data, this should easily be accomplished by a two-person team in 10 to 15 working days (to allow for inclement weather, as data recording requires meticulous work difficult in pouring rain).  The initial sampling, requiring more effort (i.e. precise mapping), is estimated to take the sampling team from 15 to 20 working days.  These estimates are based on the assumption that the sampling team can easily locate each study population.  Either they have precise location maps, or are met on site each day by local experts or staff and led to the study population.  This needs to be done only once, as the sampling team should carefully document the location of each study population (GPS data if possible, or precise position on large scale map).  A state could choose to have two sampling teams in the field at the same time, each would share in monitoring population dynamics in some populations, but could also share in the more extensive population size and structure monitoring (see section 6.2.).  The justification of two teams is to be able to visit monitored populations at an optimal time during the growing season.  The optimal time is when the fruit are still partly green and turning red.  This optimal time will vary from state to state, due to normal climate differences, and from year to year, due to annual variations in climate.  However, in general and over most of the species’ range, July is certainly the best month to conduct field work on goldenseal.  Earlier, the fruit may be under-developed and the seed they contain impossible to count by visual inspection.  Later, the fruit may have begun falling and smaller plants may have started their fall senescence.  Even by July, it is possible that small or weak plants may have already senesced, especially during a hot and dry summer.  Drought conditions are known to cause early senescence in goldenseal.  One or two populations could be visited earlier in the summer to quantify this phenomenon.  However, the overall impact of this less accurate data on population level results is expected to be slight, and a single mid-summer (July) visit to each study population is sufficient.

Populations selected for study must meet a maximum of the following criteria:

  1. A minimum number of individual plants, with 2 leaves, of 100.  This requirement can be flexible, as it is possible that such “large” populations may be very difficult to find in many areas.  Similarly, if the population contains several hundred individuals (or even thousands), efforts should be made to sample at least 250 two-leaved plants.  The idea here is to have a population which is of the Minimum Viable Population size (or as close as possible to it) determined for American ginseng, the only herb occurring in the same habitat as goldenseal for which we have MVP size data.  However, this number is only a guide, as demographic analysis will be necessary to obtain a MVP size number for goldenseal.
  2. No evidence of harvesting.  Here again, such a criteria may be impossible to attain.  Also, the population dynamics of harvested populations may be interesting.  This must be balanced against the time and money expenditures required to monitor the populations.  A population which is wiped out by harvesting before it can be sampled again (for a second year at least) will yield no data and will have wasted both time and money.  However, if a population persists for several years before being harvested, and if some plants survive the harvesting, some interesting data may be acquired.  Monitoring can then be continued to see at which time the harvested population will recover, or if it is harvested again and disappears.  The idea behind studying protected and unharvested populations is to assess how populations are growing in the absence of harvest.  Harvesting impacts can be simulated using data from these populations.  However, experience shows that few populations, even those in national parks, are secure from harvesting by poachers.  Therefore, well protected (or secluded) populations should be selected in majority for intensive population dynamics monitoring, with the understanding that some of these are likely to suffer partial or total harvesting at some point (hopefully after some useful data has been collected).
  3. Remote location (distance from road or trail).  This criteria is designed to reduce the chances that study populations will be harvested.  This must of course be balanced against the need for accessibility by the sampling teams.  A team should be able to spend 4 to 5 hours working in each population (less than half that time for the low intensity monitoring).  A 2.5 hour drive/hike to reach a population will translate into a 10 hour day (including work on site for 5 hours).
  4. Similar and typical habitat (site characteristics and vegetation).  The idea here is to avoid untypical habitats, where goldenseal may be less productive or overly so.  This factor will be of less importance as the extensive monitoring data are acquired, so that the typical or atypical nature of each intensively monitored population can be assessed by comparing it to many other populations in terms of its environmental features.  Managers or agents may wish to sample a breath of habitats in which they know goldenseal to occur in their jurisdiction.  The detailed monitoring will reveal differences in the potential of each site to sustain goldenseal harvesting or not.
  5. No evidence of recent natural disturbance (wind, herbivore, etc.).  Again, this criteria is aimed at achieving a certain sampling uniformity, but it may be interesting to include goldenseal populations that occur in young successional stands, or in forests that have been selectively cut.  Signs of herbivore pressure should be recorded for each goldenseal plant monitored, although herbivory by deer has not been reported (probably because of the strong alkaloids also found in the leaves).

6.3.3. Study population site and vegetation description

For each studied population, a site description and a vegetation description should be produced.  The data needed for these descriptions can be collected on two field data forms, the “Site and ecological data” sheet (Appendix 1) and the “Vegetation data” sheet (Appendix 2).  The first data form is used for recording all site and ecological data measured or estimated in the field.  The information it contains is completed with the analysis of a soil sample.  This soil sample, of approximately 500 ml in volume, is taken from a 5 to 15 cm depth in the mineral soil (after removing litter).  This depth corresponds to the average rooting depth of mature goldenseal plants.

Soil samples will be air dried, and passed through a 2 mm sieve before analysis.  The soil analysis should include: pH, % organic matter (loss on ignition), NO3, NH4, available P, Ca, Mg, K and Mn.  Textural analysis (soil particle size distribution) is also useful.

The vegetation data collected will consist of the estimation of the total percent cover of each vegetation strata (tree, shrub, herb, moss), and of the estimated individual percent cover for tree species, shrub species and herb species.  No specific limits will be laid out for these percent cover visual estimates, but the estimate should be representative of the vegetation associated with the goldenseal study population.

6.3.4. Demographic data sampling Locating individual plants

In each studied goldenseal population, we should be aiming at precisely locating a minimum of 100 plants with two leaves or more (more in large populations).  This number is to insure that each studied population is above the minimum viable population size, established at 172 plants for American ginseng (including first year seedlings and one-leaved plants) for populations at the northern limit of the species’ range (Nantel et al. 1996).  This number is proposed in view of the total lack of populational data for goldenseal, as well as the fact that American ginseng is a perennial herb which frequently co-occurs with goldenseal in the wild.  Habitat descriptions for both species are very similar.

Using recognizable permanent landmarks (boulder, large tree), patches of goldenseal plants will be mapped, using as many patches as necessary to achieve the required minimum number.  The mapping method will involve measuring and mapping a perimeter for each patch, using measuring tapes or topofil (surveyor measuring thread) and compass bearings, which will be reproduced to scale on a grid paper.  Mapping can be facilitated by temporarily placing colored plastic tapes on some trees and temporarily flagging each goldenseal plant encountered (all tapes and flags to be removed after mapping is completed).  The exact location of each goldenseal plant with two leaves or more will also be transferred on this map.  Each of these plants will be attributed a sequential number.  Alternatively, micro-plots could be located over several patches to facilitate the mapping, and subsequent re-locating of each plant.  One meter square plots can be used, with elastics creating a grid of 20 cm x 20 cm squares.  Two permanent corner pegs can be inserted in the soil (to a few cm above soil level) to allow for the precise placement of the micro-plots in subsequent years.  Iron rods are ideal to serve as pegs, as they can blend in well with the litter (rust color) and can be relocated using a metal detector.

Within each patch or micro-plot, particular care will be given to locating first year seedlings (small one leaf plants) and larger one-leaved plants (large five-lobed leaf).  Care must be exercised in order not to trample these small plants, which are difficult to locate in the understory.  Each of these will be attributed a sequential number as well, and their position mapped in relation to larger plants on the patch or individual micro-plot map.  Details of standing trees, logs or boulders should be added to the maps whenever possible in order to insure that individual plants, patches or micro-plots can be relocated as easily as possible.  As details about the size and number of leaves of each plant will be noted, relocation problems should be minimal the following year (as goldenseal plants will generally look the same from year to year).  Each year, new vegetative recruits (large one-leaved plants or small two-leaved plants) need to be mapped, numbered and the mother plant identified.  This will require some careful soil removal and replacement (see section 6.3.1 above).

An additional method of identifying individual plants would be to insert engraved aluminum nails (2 inches long) into the soil beside each plant.  Sequential numbers would be pre-engraved on the heads of these nails (with a common electric hand-held metal engraver), which could be relocated using a small metal detector.  This method was suggested by Dan Drees, who uses it in his monitoring of ginseng plants in Meramec State Park in Missouri.  This method was also used in 1998 when establishing two ginseng demographic study populations in the Great Smoky Mountains National Park (Gagnon and Rock, unpublished data).  However, this method is more appropriate when individual plants are scattered over the forest floor, as in the case of ginseng.  Plants mapped within micro-plots (i.e. Allium tricoccum, Nault and Gagnon 1993; Helianthus divaricatus, Nantel and Gagnon 1999) do not require this as the two corner pegs are sufficient.  Furthermore, the high density of goldenseal patches may make the engraved nail method impractical and even damaging to the plants’ roots. Data collected for individual plants

It is crucial that the data collected from each individual plant be as accurate as possible, as it will be measured again on each individual plant for many years, and will be the basis of all population dynamics analyses.  After each individual plant’s location is mapped, its demographic data in recorded on a field form (“goldenseal data”, Appendix 3).

The information recorded for each plant includes (see Figure 1): Individual sequential number, (1) height in cm to base of largest basal leaf (if only basal leaves are present) or the largest (lowest) stem leaf, (2) width (maximum) in cm of each leaf (leaf one, two and three if present) and percent of leaf area browsed, (3) the number of fruit “sections” (each is assumed to contain a seed).  For some plants with very large fruit, these counts can be difficult.  Marking each fruit section with the tip of a marker helps in keeping track as they are counted.  Such plants are usually few in each population. Sampling date and frequency, and study duration

The studied populations should be revisited in one year’s time, in early to mid-July, to record demographic data for all mapped and numbered individual goldenseal plants.  The same will be repeated in subsequent years.

Each year, extreme care will be taken to locate and record any new recruits to the population (vegetative recruits or first year seedlings).  These recruits will be numbered and incorporated in the monitoring.  Also, the mortality of individuals must be recorded, after a thorough search, and preferably including the locating of remaining dead structures (i.e. dead rhizome).  This is particularly important because of the possibility that some individuals may lie dormant for a year (perhaps following a trauma).  Objective recorded evidence of this dormancy phenomenon is presently lacking.  Goldenseal growth patterns involve producing a new annual stem on a horizontal rhizome.  This will lead to the movement of marked individuals (a small distance every year) in the forest understory.  Mapped individual plant positions may need to be updated every year or second year. Goldenseal is known to mostly propagate vegetatively.  These new vegetative recruits will need to be mapped, their demographic parameters measured, and the plant of origin (mother plant) identified (close proximity of a large plant may be sufficient, although a shallow excavation of root connections confirms origin).

6.3.5. Field and garden seed germination trials

Ideally, seed germination rates should be estimated for each study population.  In each studied population, 75 healthy seeds should be collected from large plants that are not being monitored (outside of mapped patches or micro-plots).  Seeds should be removed from the fruit pulp and washed.  A group of 50 seeds should be sown in situ in order to measure percent seedling emergence (next summer) in each population.  This information is important for the demographic analysis.  The seeds should be sown at 1 cm depth in the mineral soil and covered with a thin layer of leaf litter.  A 20 cm by 25 cm wire cage should be put on top of each seed germination trial (one per studied population), in order to prevent seed predation by small mammals.  Stiff wire pegs should be inserted through the cage and into the soil in order to firmly anchor it.

The remaining 25 seeds per population should be taken to headquarters to be placed in moist sand, buried 1 cm deep in a shaded outdoor garden where the soil is kept moist.  A wire cage, extending down into the soil (10 cm) and covering the planted seeds, should be installed to exclude vermin.  In November, 5 cm of leaf litter should be placed on top of the soil for the winter, and removed in spring (March or April).  Appearance of seedlings should be monitored in spring from April to May.

Seed germination and seedling emergence data from a controlled environment are important in accurately assessing natural seed mortality rates.  However, the taking of seeds implies that a good seed crop has occurred, and that unsampled plants with seeds are available nearby.  The seed of monitored plants should not be harvested, as this will affect future seedling recruitment in the study population.  Therefore, these experiments may not be possible in all study populations, or in some particular years (i.e. poor seed crop year).  Also, it is not necessary to repeat these experiments every year.

6.4. Overview of demographic analysis methods

Demographic analysis in plants (as in any group of organisms) is based on data recording the fate of individuals.  It is therefore extremely important to identify individuals precisely, at each time step (July to July), in order to record their demographic state at that time.  This is the reason for mapping the precise location of each studied individual plant within each study population.

Demographic fates are thus recorded each year for each individual.  First, we verify that the individual has survived, or if it has died.  Size is then measured to assess growth, lack of growth (stasis), or decrease in size (a possibility in plants).  Finally, reproductive output is measured for each individual.  This is simply measured by the number of seeds produced.  Other variables, such as number of flowers and fruits, and the cause of a low fruit crop (if evident), are not strictly necessary for demographic analysis, but are useful in identifying the causes of low seed production.  Also, each year, the studied population must be searched for the appearance of seedlings or vegetative offshoots, recruiting into the population.  In the case of goldenseal, the seeds produced in the summer will germinate in the fall, and appear as seedlings in the first spring following their production.  Thus a seed crop produced in a given fall will be reflected in the following spring’s seedling crop.

In order to proceed with the demographic analysis, it is necessary to group all the individuals studied into a small number of categories, which hopefully will contain individuals that display very similar demographic characteristics.  In population studies of animals, such categories have usually been based on age (obtained through tooth sections, growth rings on fish or turtle scales, etc.).  However, apart from trees growing in temperate and cold regions (annual growth rings), most plants cannot be aged accurately.  A conceptual breakthrough was made by Harper (1977), in proposing that size was always better correlated with demographic rates in plants, as opposed to size.  This opened the field of population biology to plants, the population dynamics of which could now be studied in populations of individuals grouped in categories of size.

The boundaries of size-class categories can be determined objectively, using a computer algorithm.  Or they can be determined more subjectively, based on size or state (vegetative or flowering plant) features easily recognizable in the field, and closely related to demographic fates.  In the case of goldenseal, size-class categories can be based on the number of leaves a plant has, as well as on its sexual reproduction state (producing a flower or not).  Seedlings and young plants have one leaf, with a much smaller and distinctive leaf for seedlings.  As they grow, they can produce, two or three leaves, and eventually a flower stem.  Seeds are  put in a separate demographic class.  Thus for goldenseal, we obtain a seed category (a state class) and five size/state-classes based on the number of leaves: 0 = seedlings (one small seed-leaf), 1 = one-leaved plants (with typical lobation), 2 = two-leaved plants, 3 = three-leaved plants, 4 = flower bearing plants (regardless of leaf number).  Field study may cause a revision in these classes (i.e. class 3 = class 4, if plants with 3 leaves always have a flower stem).

The demographic analysis consists partly in comparing the various demographic rates for each size/state-class (i.e. mortality rate, growth rate, flower production, seed production, vegetative offshoot production).  However, in order to obtain a clear picture of the growth of an entire studied population, the use of size/state-classified transition matrices, or projection matrices, is recommended (see Charron and Gagnon 1991 and Nault and Gagnon 1993).  A transition matrix is made up of cells, each of which represents the probability of a certain fate occurring during the matrix’s time step (one year in this case).  The cells represent all of the transitions from one demographic state to another, or the stasis in the same state (ex. seed to seedling, class 1 to class 2, or class 3 to class 3).  Of course, some transitions are frequent, some are rare, and some others are impossible (ex. going from seedling to seed).  The number of cells in a transition matrix depends on the number of size-classes (or combination of state- [ex. seeds] and size-classes) used.  With goldenseal we should have six state/size-classes, to obtain a matrix with 36 cells (6 x 6).

The state of individuals at time t of the study (1st year of study) is represented by the size-classes positioned at the top of the matrix.  The state of individuals at time t+1 of the study (2nd year of study) is represented by the size-classes positioned at the left-hand of the matrix.  The transition matrix is separated in two halves by a diagonal of cells running from the top left-hand corner to the bottom right-hand corner.  This diagonal is made up of “stasis” cells, for individuals which remain in the same size-class during the time step represented by the matrix.  In some cases, stasis is not allowed by the matrix; all seedlings (class 0) move on to size-class 1 or die, and we assume that all seeds germinate or die (no persistent seed bank).  In the lower half of the matrix, below the diagonal, the cells represent growth transitions (passage from size-class 1 to 2, or size-class 1 to 3, etc.).  In the upper half of the matrix, above the diagonal, the cells represent a decrease in size transitions (passage from size-class 3 to 2), which are not frequent, but still possible.

The top row of cells of the matrix represents the average seed contribution per individual for each of the size/state-classes.  The sum of all transition probabilities in each goldenseal matrix column, omitting the seed number cell, is equal to one when the mortality probability is added.  The mortality probability for each size-class is sometimes added as an independent row at the bottom of transition matrices.  If it is not directly written, it can be easily calculated as the difference between one (1) and the sum of all transition probabilities in a given column.  A “right eigenvector” of the matrix is also sometimes displayed as a column on the right-hand side of the matrix.  This represents the stable size distribution vector of the matrix.  By multiplying each number associated to a size-class in this vector by the total number of plants in the population, we can obtain the number of plants in each state- or size-class at stable size distribution.

A transition matrix is extremely easy to produce, once the difficult task of acquiring the field data has been accomplished.  Each transition matrix represents a particular studied population, and the fate of its individuals from one year to the following year.  Therefore, each transition matrix is necessarily constructed from two years of data, data from year one (ex. 1999) and data from year two (ex. 2000).  With three years of field data, we can construct two transition matrices (1999 to 2000, and 2000 to 2001) for each studied population.  To calculate the transition probability cells within a matrix, one has to proceed column by column, using the total number of individuals in the size-class corresponding to the column in year “one” (time t), for example 50 individual plants of size-class 2 in 1999, in order to divide the numbers of these same individuals in each size-class or state (i.e. death) they have transited to in year “two” (time t+1, or 2000).  For example, of the 50 size-class 2 individuals of 1999, 35 may have remained size-class 2 plants in 2000 (35/50 = 0.70), 11 may have grown to size-class 3 plants (11/50 = 0.22), and 4 may have died (4/50 = 0.08).  These transition probabilities complete the column of size-class 2 (0.70 in the 2-2 transition cell; 0.22 in the 2-3 transition cell; all other cells are 0.00; mortality rate is 0.08).  The seed production for that column is the average seed production for 1999 of individuals plants in size-class 2 (ex.  56 seeds produced / 50 size-class 2 plants = average of 1.12 seeds/plant).

Several useful demographic interpretations can be made directly from the study of individual matrices.  For example, decreasing mortality can be observed as plants become larger, or increase in size-class.  Stasis (remaining in the same size-class) can also be observed to be the most prevalent transition, particularly in larger size-classes (2 and higher).  However, the most useful single piece of information that can be obtained from a transition matrix, is the growth rate (l) of the population it represents.  The growth rate indicates if a population is stable (recruitment equal to mortality, l = 1.0), or expanding (l ³ 1.0), or decreasing (l ² 1.0).  For example, a growth rate of 0.95 indicates that a population is decreasing at a rate of 5 % annually.  Similarly, a growth rate of 1.10 indicates that a population is expanding at a rate of 10 % annually (increase in total number of individuals, spread over all state or size-classes, in proportion to the normal size distribution of the population).

In order to extract more information out of each transition matrix, they can also be transformed into elasticity matrices (see method in De Kroon et al.1986).  Elasticity matrices are basic transition matrices in which each cell is transformed in order to express its specific contribution to the population growth rate.  The cells with the largest values in an elasticity matrix indicate which transition, or transitions, contribute the most to the population growth rate.

The population growth rate obtained for each individual matrix identifies the growth trend of the population represented by each matrix, during the time step represented.  This is why transition matrices are referred to by some as projection matrices.  One can project into the future the population growth trend obtained, in order to predict if the population is stable, growing (slowly or fast) or declining (slowly or fast).  However, this prediction implies that the environmental conditions occurring during the time step represented by the matrix (one year) will remain the same in the future.  This is certainly not true, as year to year variations in environmental conditions are common and frequent (drought, frost, browsing, pathogens, etc.), and they have a significant influence on demographic parameters (flower abortion, seed production, growth, mortality, germination, seedling emergence, vegetative offshoot production).  At best, the population growth rate derived from one matrix is a good snapshot of the status of the studied population (stable, expanding, declining).

In order to provide a better tool for predicting the survival and growth of a population, several matrices are needed.  These can be used in stochastic population projections, which give a more realistic picture of the population dynamics over time.  Stochastic population projections are achieved by calculating the growth of a population using a series of matrices in a random order.  This is more similar to what occurs in a natural environment, where years of good growth will alternate with years of poor growth, and sometimes several good or bad years in a row will occur.  These different matrices should optimally come from the same population studied over several years, such as the four matrices produced for a wild leek (Allium tricoccum) population by Nault and Gagnon (1993).  However, if several populations from similar ecological conditions (i.e. same elevation, habitat and vegetation type) are studied at the same time, all of their matrices can be pooled to produce the series of matrices necessary for stochastic population projections (Nantel et al. 1996).

Stochastic population projections can be used to calculate a minimum viable population size for goldenseal, for each distinct region or state studied (Nantel et al. 1996).  Similarly, these projections can be used to predict the effects on population survival of various levels of harvesting (Nantel et al. 1996).  The method used is to produce 100 series of randomly ordered matrices, from those available (a minimum of four should be used; they can be obtained initially by studying four similar populations for two years [one transition per population], or better by eventually having as many transitions as possible for each study population).  Each random series is different, as the order of appearance of each matrix is different within each, and represents 100 years.  Minimum viable population size can be empirically determined by doing 100 population projections (using the 100 random series) for each one of different starting population sizes.  The largest population size at which 5 % or more of the series become extinct, is the minimum viable population size.  This minimum viable population size (MVP) is defined as a population size likely to give a population a 95 % probability of survival over a period of 100 years (Menges 1992).

Similarly, harvesting impacts on population survival and growth can be tested using stochastic population projections.  The appropriate cells of each matrix (i.e. those of all plants except seedlings), can be modified to account for harvesting by simply increasing mortality.  In the case of goldenseal, the harvesting of a plant is the demographic equivalent of its death; it is entirely removed from the population.  A refinement possible is the frequency of the harvesting.  It can be every year, every three years, every five years or more, or follow a random pattern (i.e. discovery by poachers).


Bierzychudek, P.  1982.  The demography of jack-in-the-pulpit, a forest perennial that  changes sex. Ecological Monographs 52: 335-351.

Bowers, H.  1891.  A contribution to the life history of Hydrastis canadensisBot. Gaz. 16: 73-82.

Catling, P.M. and Small, E.  1994.  Poorly known economic plants of Canada.  3.  Hydrastis canadensis L.  Canadian Botanical Association Bulletin 27(3): 50-51.

Charron, D. and Gagnon, D.  1991.  The demography of northern populations of Panax  quinquefolium (American Ginseng).  J. Ecol. 79: 431-445.

Davis, J.M.  1995.  Advances in Goldenseal cultivation.  North Carolina State University,  Agricultural Extension service.  Horticultural Science, Mountain Horticultural  Crops Research and Extension Center, Fletcher, North Carolina.

De Kroon, H., Plaisier, A., Van Groenendael, J. and Caswell, H.  1986.  Elasticity: the  relative contribution of demographic parameters to population growth rate. Ecology 67: 1427-1431.

Eichenberger, M.D. and Parker, G.R.  1976.  Goldenseal (Hydrastis canadensis L.)  distribution, phenology, and biomass in an oak-hickory (Quercus-Carya ovata)  forest. Ohio Journal of Science 76: 204-210.

Fetcher, N. and Shaver, G.R.  1983.  Life histories of tillers of Eriophorum vaginatum in  relation to tundra disturbance.  J. Ecol. 71: 131-147.

Harper, J.L.  1977.  Population Biology of Plants.  Academic Press, New York.

Kinoshita, E.  1987.  Sex change and population dynamics in Arisaema (Araceae) I.  Arisaema serratum (Thunb.) Schott.  Plant Species Biology 2: 15-28.

Meagher, T.R.  1982.  The population biology of Chamaelirium luteum, a dioecious member of the lily family: two-sex population projections and stable population structure.  Ecology 63: 1701- 1711.

Menges, E.S.  1990.  Population viability analysis for an endangered plant.  Conservation  Biology 4: 52-62.

Menges, E.S.  1992.  Stochastic modeling of extinction in plant populations.  In:  Conservation biology: the theory and practice of nature conservation, preservation  and management.  P.L. Fiedler and S. Jain (Eds).  pp. 253-275.  Chapman and  Hall, New York

Menges, E.S. and Gordon, D.R.  1996.  Three levels of monitoring intensity for rare plant  species. Natural Areas Journal 16: 227-237.

Moloney, K.A.  1988.  Fine-scale spatial and temporal variation in the demography of a  perennial bunchgrass. Ecology 69: 1588-1598.

Nantel, P., Gagnon, D. and Nault, A.  1996.  Population viability analysis of American  Ginseng and Wild Leek harvested in stochastic environments.  Conservation Biology 10: 608-621.

Nantel, P. and Gagnon, D. 1999.  Variability in the dynamics of northern peripheral vs  southern populations of two clonal plant species, Helianthus divaricatus and Rhus  aromaticaJ. Ecol. (in press).

Nault, A. and Gagnon, D.  1993.  Ramet demography of Allium tricoccum, a spring  ephemeral, perennial forest herb.  J. Ecol. 81: 101-119.

USFWS/OSA.  1997.  Proposal to include Hydrastis canadensis in Appendix II of the  Convention on International Trade in Endangered Species of Wild Fauna and Flora  (CITES).  10th Meeting of the Conference of the Parties, Harare, Zimbabwe.

Van Fleet, W.  1949.  Goldenseal under cultivation. USDA Farmers’ Bulletin no.613  (original published in 1914, revised in 1949 by A.F. Sievers).

Werner, P.A. and Caswell, H.  1977.  Population growth rates and age versus stage- distribution for teasel (Dipsacus sylvestris Huds.).  Ecology 58: 1103-1111.

White, D.J.  1991.  Status report on the Golden Seal, Hydrastis canadensis L., in Canada. Committee on the Status of Endangered Wildlife in Canada (COSEWIC), Ottawa.

Appendix 1: “Site and ecological data” field form


SITE NO.:_________  NAME OF SITE:_______________________________
OBSERVERS:_______________________________ DATE:_______________
AREA (m2):________           PHOTOS:_____  SOIL SAMPLE:_____
SOIL TYPE:____________ HUMUS TYPE:______ BEDROCK TYPE:____________
LATITUDE-LONG. (or map):____________________
1.    Elevation _____   m
2.    Aspect  _____   degrees
3.    Slope  _____   %
4.    Topo. position _____  1-6 1=crest




5.    Drainage _____  1 - 6 1=excessive

3=moderately good



6=very poor
6.    Flooding _____  1 - 6 1=never
2=rare, occasion.

3=tempor.< 6 mo.

4=tempor. > 6 mo.

5=permanent shal.

6=permanent deep
7.    Outcrops _____   %
8.    Surficial material ___   1 - 6 1=bedrock




9.    Rocks in soil _____  0 - 5 0=0%




10.    Soil depth _____   cm
11.    Charcoal _____  0 - 3
12.    Mottles  _____  0 - 3 0=absent


13.    Watertable _____  0 - 1 0=below 30 cm
1=above 30 cm
14.    Light  _____  1 - 3 1=open

15.    Seepage _____  Y / N
16.    Sketch and comments


Appendix 2: “Vegetation data” field form


SITE NO.:_________  NAME OF SITE:_______________________________
OBSERVERS:_______________________________ DATE:_______________

VEGETATION TYPE:__________________________________________________

SUCCESSIONAL STAGE:______________________________________________

% COVER OF STRATA TREE:____  SHRUB:____  HERB:____  MOSS:_____

SPECIES and % COVER CLASS (1 = 0-1%; 2 = 1-5; 3 = 5-25; 4 = 25-50; 5 = 50-75; 6 = 75-100)

Appendix 3: "Goldenseal data" field form


SITE NO.:____ NAME:_____________ OBSERVERS:___________DATE:__________
Size/state class

Height to base of largest leaf (basal if only basal present, or lowest stem leaf),

Width (max.) of each leaf (indicate % browse),

Seeds (count fruit sections),

Vegetative connection (no. of mother plant)