McGraw, Jim. Eberly Professor of Biology, West Virginia University.

(Presented at The Future of Ginseng and Forest Botanicals Symposium, July 12-14, 2017, Morgantown, WV)

Abstract

American ginseng has been harvested from the wild to supply the Asian traditional medicine trade in North America since the early 1700s.  However, only since 1975 have federal and state regulations been in place to regulate this trade, and the harvest practices supporting it.  Using data from a unique long-term formal census of 30 natural populations in 7 states, my lab examined recent trends from 2004-2014 as an indicator of how wild ginseng populations are faring.  Combined with demographic modeling ‘experiments’ using these census data, my students have also examined alternative projections for the future of ginseng over coming decades, including factors that explain variation in population growth and viability.  Over the decade (2004-2014), 25 of 30 ginseng populations declined in size from their initial numbers.  The mean decline over this timeframe was 30%.  The observed immediate reasons for the decline were (1) harvest; particularly illegal harvest, which was prevalent in about half the populations, and (2) deer browse; also prevalent, and distributed unevenly among populations.  Age-specific demographic data clearly showed that the current 5-year age restriction is insufficient to protect wild populations.  Additional analyses showed large between population variation in age-specific growth and reproduction, suggesting that an alternative criterion for harvest is needed.  Also negatively impacting populations were hot, dry summers that curtailed growth and seed production, as well as intense natural or human-caused tree canopy disruption.  In the long-term, landscape level changes to the forest, combined with climate change, are projected to impact ginseng populations negatively in the coming century.  While regulations imposed in response to CITES listing may only impact harvest practices, they must be considered in light of the broader array of stresses experienced by natural populations.  The precarious state of wild populations even in the absence of harvest pressure suggests that wild harvest practices need to be further adjusted toward the ‘stewardship’ end of the harvest spectrum.  At the same time, forest ‘farming’ practices using regional ginseng ecotypes that relieve harvest pressure on wild populations need to be encouraged through any regulatory reform that takes place.

 Introduction

The most important question concerning wild American ginseng populations is ‘Are populations declining?’ Historical data suggests that certainly individuals in the past were much larger than individuals today.  We can see that from historical photographs (Fig. 1).  We can also see the same phenomenon when studying herbarium specimens.  Herbarium specimens from the 1800s and early 1900s were much larger than herbarium specimens today (McGraw 2001). Other indirect evidence includes anecdotes about fortunes made from the harvest of ginseng in the 1700s and 1800s. This includes large shipments that comprised John Jacob Astor’s fortune and shipments made by Daniel Boone Down the Ohio River.  We can also hear anecdotes from harvesters today about how ginseng is getting harder to find.

One more scientific case study of ginseng harvest also involved herbarium specimens collected at different points in time (Case et al. 2007).   This study showed that at least for part of ginseng’s range the rate of collection of specimens by botanists has slowed down relative to rates of collection earlier in the history of herbarium collections.

In Search of Direct Evidence for Population Growth or Decline

Ideally, we would prefer direct evidence for decline in ginseng populations but the kind of sampling required to obtain direct evidence provides unique challenges. One of these challenges is that there are in fact tens of thousands of small ginseng populations arrayed over a wide area of the eastern deciduous forest of the United States.  Random sampling of such a large area would involve assigning a random number to each of these small ginseng populations, sorting by the random number, then choosing approximately the top 1000 to visit and follow over time.  This statistically robust sampling scheme is impractical in the extreme.  However, every practical sampling scheme will end up being either slightly biased or inadequate in some measure.  To gauge natural population status, we therefore attempted a more modest goal: to establish a representative subset of populations across a wide range. Henceforth I will call these ‘NSF LTREB populations’ after the National Science Foundation program that funded the project (McGraw et al. 2017).  By 2004, we had established 30 populations located in seven states for our sampling, including three states (WV, KY, VA) that are high harvest states (Fig. 2). These populations occurred across a wide array of elevations (397 – 3504 feet). They also occurred across a wide range of pH levels (3.9 – 6.6). In addition, these populations were found over a wide range of soil types, overstory, land ownership, land use history, and access. In all, we are acquired 457 ‘population years’ of data.  Each year, we sampled between 4,300 and 5,200 plants.

In each population, censusing in year 1 consisted of careful surveys to locate plants.  Plants were considered to be in a different population if they were at least 100 m from any other such grouping.  This distance was based upon previous studies of pollination and the drop off and seed production as a function of distance from a pollen source (Hackney, 1999).  Annual censusing thereafter consisted of visiting populations in spring and fall of each year. The spring census occurred between May 20 and June 20 while the fall census occurred between August 1 and August 20.  The spring census was used to measure germination, leaf lengths/widths (and infer leaf area from those using multiple regression), plant height and several other conditions of fully expanded plants. The late summer census was used to assess changes in status over the growing season and to count seeds on every plant in every population (Fig. 3).

In order to summarize these data, we used a simple general equation for population change:

(1) N_t+1 = N_t + births – deaths + immigration

This was simplified to exclude immigration because this term was assumed to be so small as to be negligible.  ‘Births’ were counted when new seedlings arose by germination, while deaths were designated when plants disappeared from the population for two years or more.

(2) N_t+1 = N_t + new seedlings – deaths

For presentation, number changes over time were relativized to an initial population size of 100 for all populations, and the starting year was 2004, the first year for which we had data on all 30 populations.  The first year of population change was therefore:

(3) N_{relative, 2005} = 100 * N₂₀₀₅/N₂₀₀₄

Subsequent years were given by the general equation:

(4) N_{relative, t+1} = N_{relative, t} * N_t+1/N_t

Equation 4 was iterated up to 2014, after which determination of deaths was not possible because not enough years had passed to definitively determine mortality.

Figure 4 shows the population trajectories from 2004 through 2014.  Significantly more than half of the populations declined over this time frame (G=25.49, p<.0001).  Five populations increased relative to their 2004 values while 25 populations decreased in size. The mean decline over the decade was 30%, while the median decrease was 39%.  Some years showed very little decline, e.g., 2007 – 2008 and 2009 – 2010, however most years exhibited a decline in the majority of populations.  The fact that five populations actually increased over the decade suggests that ginseng decline is not inevitable, and these populations represent an opportunity to explore the differences between increasing and decreasing populations.

The Role of Harvest in Population Decline

McGraw et al. (2013) suggested that one of the main reasons for population decline was harvest. In one early study of harvest Van der Voort and McGraw (2006) showed that the behavior of harvesters had a significant negative effect on population numbers.  Van der Voort et al. (2006) simulated population growth as affected by three harvester behavior types: (a) non-compliant (with existing harvest regulations), (b) compliant, and (c) stewardship harvesters.  The most important differences between these are highlighted in Table 1.

Table 1

These differences in harvester behavior had a significant and important effect on population growth (Van der Voort and McGraw 2006).  Relative to unharvested populations, only the stewardship harvest had equivalent population growth (Fig. 5). Non-compliant harvest reduced population growth 15% on an annual basis, while even compliant harvest reduced growth 8%.  Populations subject to both non-compliant and compliant harvest would soon go extinct.  These results showed that movement of harvest policy toward more compliant behaviors was not sufficient to prevent extinction.  Instead, harvest behavior needed to be closer to stewardship in order to be sustainable.

A subsequent analysis of actual harvester behavior in harvest events observed in LTREB populations showed that most harvest events were in fact illegal in one of three respects (Fig. 6) (McGraw et al. 2010).  Two of these (noncompliance with harvest season, harvest of undersized plants) were shown by Van der Voort and McGraw (2006) to have strong negative effects on population growth rate (Table 1).  This rate of noncompliance with existing regulations suggests that solutions to the problem of ginseng conservation need to consider how to regulate in such a way as to improve compliance.  New regulations may be well-intended, and biologically sound, but if compliance is lacking, they will be ineffective.

In the early 2000’s, harvest seasons varied widely among states with no apparent rationale based on geography or biology (Fig. 7).  However, the finding that across the range seeds were not ripe for dispersal on Aug. 15 (McGraw et al. 2005) led to convergence of harvest season to Sept. 1 (Fig. 7).  This policy should, in theory, move harvest more toward a stewardship relationship of harvester to the plant, though this change will be effective only in so far as harvesters comply with the new regulation.  For example, a severe harvest was observed June 7, 2017 in Population 29 (of the NSF LTREB populations); one that is sure to have negative consequences as a large portion of the seed-producing plants were taken (Fig. 8.).

The ‘Five-Year Rule’ and Sustainable Harvest

Many opportunities for revising harvest regulations more extensively could improve prospects for sustainability.  One consistent regulation among states involves the so called ‘5-year rule’, whereby US FWS stipulates that plants must be five years of age or older to be exported.  Using the LTREB data set, in which individual plants have been followed from ‘birth’ (germination) allows new analyses of age-specific life histories to evaluate how well the 5-year rule protects populations.  Many misconceptions exist regarding the rate at which plants survive, grow, and reproduce with age.  Using thousands of new seedlings followed through time, the analysis shows that by age 5, plants may survive at a rate of 90% annually, but on average they only produce 0.6 seeds each (Fig. 9).

Age-specific reproductive and survival schedules do not tell us directly how many seeds would need to be required for a new germinant to replace itself.  To estimate that, we can examine the fraction of seedlings remaining alive by age 5 (Fig. 10).  In five western LTREB populations (IN and w. KY), only 25% of new germinants remain after 5 years, suggesting that at least four seeds would need to be produced by that age.  This number would be 3 seeds in north central populations.  However both of these figures do not account for losses of seeds from the seed bank.

How old do plants have to be before they have literally ‘replaced themselves’ after germination?  This can be determined by performing actual simulations of a cohort of 100 new germinants, then determining how long it takes for this cohort to produce 100 new germinants. One subtlety in making this calculation is the determination of new recruitment.  In censusing ginseng populations, we carefully assessed the area for new recruits within 2 meters of each plant.  However, this may not include all the new recruits since some dispersal may occur beyond 2 meters.  If we make an ‘optimistic’ projection of our under-censusing of new seedlings, we find that it takes 12 years for a single plant to replace itself.  This optimism is based on a fairly limited observation of dispersal beyond 2 m by birds (specifically wood thrushes; Hruska et al. 2014, Elza et al. 2016).  Being pessimistic about dispersal by birds, we find that it may take as much as 22 years for a plant to replace itself (Fig. 11).  The actual replacement age is probably somewhere between age 12 and 22.

Further convincing evidence that age 5 as a harvest threshold is not protecting the population from harvest is the low frequency of flowering (Fig. 12) and the immaturity of plants (Fig. 13) by age 5.  Indeed, most states stipulate the plants must have three leaves or more to be harvested, but in all regions, less than 1/3 of plants have reached the 3-leaf stage by age 5.

Alternatives to the Five-Year Rule

As a solution to the inadequacy of the 5-year rule, it would be tempting to choose a new (greater) age limit to protect populations.  Indeed one advantage of age as a criterion is that it is verifiable at the points of sale and export.  However, the plasticity of plants as a function of age means the choice of any age is likely to be protective in some populations but not in others.  This can be seen readily from photographs of plants in different age classes in two populations (Fig. 14).  These two populations are comparable in overall size, but one grows in a moderately supportive habitat, while the other is a near-ideal habitat.  In Population 27, plants are frequently deer browsed, and have other unknown stresses that cause them to grow slowly.  In Population 30, plant progress rapidly through the stage classes and reach 3-leaf reproductive size by age 8 quite often.  Harvest of age 5 and above plants in Population 27 would cause a rapid crash because there would be virtually no seed production there, however some plants would produce seeds at age 8 and above in Population 30 (though harvest at age 5 would still be strongly inhibiting).

By contrast to the age-specific growth patterns in Population 27 and 30, the size-specific patterns are similar (Fig. 15).  This suggests that a size-based criterion for harvest could protect these two populations well, while an age-based criterion would leave Population 27 vulnerable.

To regulate harvest according to leaf area would be unrealistic since harvesters cannot be expected to measure leaf lengths and widths, then use multiple regression to compute leaf area, all while hiking in remote back country!  This opens the question, however, of what size-based criterion could be used to good effect, while allowing it to be usable by harvesters, verifiable (preferably at both point of sale and export), and demographically meaningful.  It would be tempting to suggest some root size criterion, expect that root harvest is destructive (although planting roots back after excavation is possible, it does cause some losses).  Aboveground size dimensions would be possible, but would require tops to be turned in along with roots.  If tops have some economic value, this could incentivize this behavior and perhaps also discourage out of season harvest.  These are a few of the factors that should be considered in a move to a size-based harvest and export criterion.

The Larger Context

Harvest is one of many stressors acting on ginseng populations to reduce population growth (McGraw et al. 2013; Fig. 16).  Two other factors – deer (McGraw and Furedi 2005) and climate change (Souther and McGraw 2011a, b; Souther et al. 2012; Souther and McGraw 2014) have effects of comparable magnitude, and in specific populations other factors can further depress population growth either chronically or episodically (McGraw et al. 2013).  If harvest was acting alone, more harvest could occur before population growth would be depressed below 1 to the declining level (Fig. 16), however in the presence of these factors, populations sit precariously close to that tipping point.

Summary

The challenge of the ‘Ginseng and Forest Botanicals’ Conference is to identify specific improvements we can implement in coming years to move toward ‘stewardship’ harvest from the current state of ‘exploitative’ harvest of vulnerable populations.  This challenge is not easy as it involves multiple stakeholders with competing interests.  Nevertheless, meeting this challenge will be necessary if we are to guarantee the persistence of wild ginseng populations for future generations.  Shifting our policy goals from the short-term and pragmatic toward goals that are optimistic, idealistic and long-term, can be a driving force for making the necessary changes to the human-ginseng relationship that will preserve that relationship.  Purposeful evolution of the human-ginseng relationship toward that of a mutualism, rather than a parasitic or predatory one, may be one measure of our maturity as a species.

Acknowledgments

First and foremost, the author wishes to thank the series of hardworking graduate students who led research teams each summer to do the hard work of data collection that led to the publications cited in this paper.  These included: Martha Van der Voort, Mary Ann Furedi, Emily Mooney, Zach Bradford, Kerry Wixted, Sara Souther, Jessica Turner, and Jennifer Chandler.  Colleague Anne Lubbers at Centre College was instrumental in censusing Kentucky populations.  Emily Thyroff played a key role in curating the data set resulting from this study.  Several dozen undergraduate conservation interns also assisted in the hard work of data collection, data entry and data checking.  This research was supported by a series of grants from the National Science Foundation (DEB-0212411, DEB-0613611, DEB-0909862, DEB-1118702).  Finally, I would like to acknowledge the contributions from my family, who tolerated my summer absences during ginseng census seasons so I could do my part to maintain the consistency of the data collection effort.

References

Case, M. A., K. M. Flinn, J. Jancaitis, A. Alley, and A. Paxton. 2007.  Declining abundance of American ginseng (Panax quinquefolius L. )  documented from herbarium specimens.  Biological Conservation 134: 22-30.

Elza, M. C., C. Slover, and J. B. McGraw. 2016.  Analysis of wood thrush (Hylocichla mustelina) movement patterns to explain spatial structure of American ginseng (Panax quinquefolius) populations.  Ecological Research 31: 195-201.  DOI 10.1007/s11284-015-1327-6.

Hackney, E. E. 1999. The effects of small population size, breedingsystem, and gene flow on fruit and seed production in Americanginseng (Panax quinquefolius L., Araliaceae). MS thesis. West Virginia University, Morgantown.

Hruska, A. M., S. Souther, and J. B. McGraw.  2014.  Songbird dispersal of American ginseng (Panax quinquefoliusL.). Ecoscience 21(1):1-10.

McGraw, J. B. 2001. Evidence for decline in stature of American ginseng plants from herbarium specimens.  Biological Conservation 98: 25- 32.

McGraw, J. B., M. A. Furedi, K. Maiers, C. Carroll, G. Kauffman, A. Lubbers, J. Wolf, R. Anderson, R. Anderson, B. Wilcox, D. Drees, M. E. Van der Voort, M. Albrecht, A. Nault, H. MacCulloch, and A. Gibbs.  2005.  Berry ripening and harvest season in wild American ginseng.  Northeastern Naturalist 12(2): 141-152.

McGraw, J. B. and M. A. Furedi.  2005.  Deer browsing and population viability of a forest understory plant.  Science 307: 920-922.

McGraw, J. B., S. Souther, and A. E. Lubbers.  2010. Rates of harvest and compliance with regulations in natural populations of American ginseng (Panax quinquefolius L.).  Natural Areas Journal 30: 202-210.

McGraw, J. B., A. E. Lubbers, M. E. Van der Voort, E. H. Mooney, M. A. Furedi, S. Souther, J. B. Turner, J. Chandler. 2013.  Ecology and conservation of ginseng (Panax quinquefolius) in a changing world.  Annals of the New York Academy of Sciences 1286: 62-91.  {ISSN 0077-8923.  DOI: 10.1111/nyas.12032.  (Invited Review)}

McGraw, J. B., M. E. Van der Voort, Mary Ann Furedi, A. E. Lubbers, E. Mooney, S. Souther, J. Turner-Skoff, J. L. Chandler, E. Thyroff, 2017.  Long Term Research in Environmental Biology: Demographic census data for thirty natural populations of American Ginseng: 1998-2016. Long Term Ecological Research Network.  http://dx.doi.org/10.6073/pasta/d563972ac7e6a14a9b34bdad08d18e3a

Souther, S. and J. B. McGraw. 2011a.  Vulnerability of wild American ginseng to an extreme early spring temperature fluctuation.  Population Ecology 53(1):119-129.

Souther, S. and J. B. McGraw.  2011b. Local adaptation to temperature and its implications for species conservation in a changing climate.  Conservation Biology 25(5): 922-931.

Souther, S., M. J. Lechowicz, and J. B. McGraw.  2012.  Experimental test for adaptive differentiation of ginseng populations reveals complex response to temperature.  Annals of Botany 110: 829-837.

Souther, S. and J. B. McGraw.  2014.  Synergistic effects of climate change and harvest on extinction risk of American ginseng.  Ecological Applications 24(6): 1463-1477.

Van der Voort, M. E. and J. B. McGraw.  2006.  Effects of harvester behavior on population growth rate affects sustainability of ginseng trade.  Biological Conservation 130: 505-516.