The Gilded Age Garden Hypothesis

How, When & Where was Hemlock Woolly Adelgid Introduced to Eastern North America?

Not much attention has been given to the historical introduction of HWA to the Eastern US. We know that HWA was first identified from a 1951 sample collected in Richmond VA. But was Richmond the only HWA introduction site? And was that introduction really as late as 1951?

There are two important lines of scientific research on HWA that can help address this issue. One line of research has applied analysis of mitochondrial DNA to all known Hemlock Woolly Adelgid populations in the world, to identify the place of origin of the HWA introduced to the eastern US. The other area of scientific research has produced a historical map, utilizing USDA’s historical HWA detection records, that can be used to examine the HWA introduction process for the eastern US. And combining these two resources provides a scientific foundation for examining the HWA introduction process.

The HWA genetic research published by Nathan Havill and associates (in 2006 & 2016) has transformed our understanding of world HWA diversity! This research has not only documented biological differences between HWA lineages in the US and multiple Asian countries. It also has definitively identified the HWA found in the Eastern US as originating in Southern Japan. Below is the 2006 chart identifying the biologically distinct HWA lineages. (The 2016 analysis identified three additional HWA lineages located in China, Taiwan and Ullung Island.)

Where was Hemlock Woolly Adelgid Introduced?

This DNA analysis tells us that “our” HWA came from Southern Japan, but how did this introduction occur – at a single site or at multiple locations in the Eastern US? We can use USDA historical data on HWA identification sites during the 30+ year period from 1971-2002 to begin to address this question. But please note that these USDA-collected data do not represent dates for the “First Introduction” of HWA (as labeled), instead they are dates for the “First Observation” of HWA by USDA representatives in the mid-20th century.

The map below uses county-level data, prepared by the US Forest Service, representing the year in which HWA was confirmed present in each county in the Eastern US. Note that the sites of earliest HWA observations (1971-1984) are marked with the darkest red color, the next set of HWA observations (1985-1990) are marked a tan color. And later HWA observation sites (1991-2002) are marked bright red.

To interpret these data, we can ask: Where are the early HWA introduction sites in the Eastern North America? And what is the pattern of HWA movement over time? For the first question we can note 5 geographically distinct sites where early HWA infestations were reported, as indicated by dark red coloring for Central Virginia mountains, Richmond, coastal Virginia, Philadelphia and Long Island. For the latter question we need to note the transition from dark red areas to tan areas to bright red areas, as the HWA are observed moving out to counties with previously uninfested hemlock areas.

My objective here is to use these historical data on HWA distribution and dispersion to “test hypotheses” about single vs multiple HWA introductions. And again a cautionary reminder that these data do not represent dates for the “First Introduction” of HWA (as labeled), instead they are dates for the “First Observation” of HWA. And I will provide evidence that the actual HWA introduction dates are much earlier than those presented in the map.

The first map below depicts the single-HWA-introduction-in-Richmond hypothesis which has dominated most USDA interpretations. Here, the outward distribution from Richmond works reasonably well in accounting for HWA outflow in the Southeast, but it does not account well for the observed HWA distributions in the northeastern US.

Single Site HWA Introduction Hypothesis

The second map below presents the competing multiple-HWA-introduction-sites hypothesis, using four early HWA introduction sites suggested by the USFS data: Central Virginia mountains, Richmond, Philadelphia and Long Island.

Multiple Site HWA Introduction Hypothesis

This multi-site hypothesis seems to work much better in accounting for HWA dispersion throughout the eastern US – north as well as south. But what could be the explanation for specific HWA introduction sites proposed?

How and When Was Hemlock Woolly Adelgid Introduced?

This question was first raised in 2008 by USDA researchers Nathan Havill and Michael Montgomery, as they searched for a “Japanese connection” in historical documents for Maymont Estate in Richmond, Virginia – the site of the original HWA discovery. And this search led to reports on a 1911 Japanese Garden construction project at Maymont, involving international garden designer Y Muto.

HWA biological research suggests that cross-continental HWA transfers would require “host plants” from one of two species of live plants native to southern Japan, the Southern Japanese Hemlock (Tsuga seiboldii) or the Tigertail Spruce (Picea torano). And Muto’s affiliation with the world-famous Yokohama Nursery provides an international commercial source for such Japan-origin landscape plants.

More recently, Patrick Horan (2017) has conducted an expanded search for additional Muto-affiliated ‘Gilded Age Garden’ sites in the eastern US. This search has located documentary evidence for three more early 20th century gardens involving Mr. Muto – in addition to Richmond. And this provides a Muto-affiliated ‘Gilded Age Garden’ site for each of the locations identified in the USFS map as possible early HWA introduction sites.

While this evidence for Muto-affiliated garden sites in Long Island, Philadelphia, Richmond and the central Virginia mountains is suggestive, it doesn’t prove that HWA was introduced at these sites. However, it does provide a plausible hypothesis that is consistent with both USFS data and existing archival documentary evidence for these early 20th century garden sites. (This evidence includes Muto’s garden involvement, the presence of Muto-era Japanese Hemlock and Spruce trees at these sites and documentation of purchases of same from the Yokohama Nursery.)

Gilded Age Garden Sites

The map above presents the four Muto-affiliated Gilded Age Garden sites that have been identified to date. Three of these sites have some archival evidence as well as the involvement of internationally-oriented garden owners: John and Lydia Morris of Compton Estate in Philadelphia, and James and Sallie Dooley at both their Maymont Winter residence in Richmond and their Swannanoa Summer residence in the central Virginia mountains.

Both John Morris and Sallie Dooley were noted for their participation in international horticulture in the early 20th century. Morris had employed Mr Muto for multiple turn-of-the-century Philadelphia area Japanese garden projects, including several on his own Compton estate. And James Dooley described his wife’s garden orientation as involving “purchase [of] the most costly evergreens from all parts of the world”.

But for Long Island, beyond a report of Muto’s 1910 participation in a Japanese garden project, there is no evidence indicating the specific site of this garden. As a large Japanese garden project pictured in 1915, Marshfield Estate “fits the profile”. But there is no further evidence connecting this site to Muto or Japan-origin plants. So this should only be considered as a “possible” Long Island candidate, pending further historical research on Long Island gardens of that era.

Finally, the  Virginia coastal site, indicated on the USFS map, is more likely a later garden HWA transfer from Richmond rather than an introduction from Japan.

Directions for Further Research

There are important differences between scientific applications to contemporary topics, as opposed to historical topics. For example, it is often possible to collect additional data to address different hypotheses that arise for contemporary topics, whereas it is rarely possible to go back in time to collect additional data to address historical topics. But here are some suggestions about possible additional historical data that would be relevant to this inquiry.

Further archival research will be needed to evaluate and elaborate the multi-site HWA introduction hypothesis proposed above. I welcome the participation of others in this research effort. And I acknowledge the helpful contributions of Curators Dale Wheary at Maymont and both Leslie Crane-Smith and Anthony Aiello at the Morris Arboretum to my own research efforts

Another valuable research extension would involve the original HWA observation reports collected by USDA, which were used to construct the USFS historical county-level map utilized here. There remain some important mysteries or unknowns about these data. Who were the USDA representatives reporting on HWA sightings and where are the resulting HWA site reports located?

My current hypothesis is that these data were collected by USDA County Extension offices in each state. And that would mean that these historical data for local HWA identifications are probably stored in USDA Extension archives at the land-grant university in the respective states. But I have not yet been able to confirm this with officials at USFS or USDA.

Patrick Horan

Saving Hemlocks