The start of this project was delayed this year (2022) for two main reasons; 1) we encountered issues with our plans to restore the creek bed before planting fodder blocks because of slowdowns working with local NRCS / Soil and Water agencies, and 2) the outbreak of the Spongy Moth (Lymantria dispar) caterpillar led to almost 100% mortality of our willow cuttings. We'd never experienced such a devastating loss, as usually cuttings are quite easy to produce in previous years. 

As a result, we asked SARE to delay our end date and allow us to just push all research and activities back one year. We have changed the date in the above description to new targets. 

We did successfully collect cuttings and develop a procedure for starting them in "air prune" beds, which were also constructed and tested. This nursery elements proved to be a marked improvement over previous methods of propagation of cuttings (in ground beds)-- while almost all of the willow and poplar suffered at the hands of the spongy moth, we had elderberry cuttings in the beds that thrived and produced impressive root systems. We do think this will allow for a good strategy to propagate willow/poplar for next year and allow for healthy root growth. 

According to pest experts, the spongy month epidemic in our area should subside next year, as the cycle of intense impact of the population explosion tends to two years, and 2022 was the second season. Even still, we will plan on extra monitoring and treatment if necessary of the beds (using Bacillus thuringiensis kurtaki (Btk) approved in Organic production)

2023 Update

For this season, we took cuttings of willow and poplar from the farm landscape and also local known cultivars of willow and unknown varieties of poplar. We planned to match this effort with the cost of purchasing cuttings, which we did from Vermont Willow Nursery (average cost of $2.90 per cutting) and hybridpoplars.com (average cost .56 per bud cutting).

Unfortunately, the ordered stock did very poorly in our rooting beds both in the field and woods, with near complete mortality, while the cuttings we took did quite well. Some combination of management from the suppliers (cuttings are often stored in a cooler or freezer for a long time?) and our management (we didn’t get everything out into the beds as quick as we’d have liked) probably led to this high mortality. The Poplar trees were received in early March but sat in our cooler (packed in wet peat) until the end of the month. The Willow was ordered in March but didn’t arrive until late April, so may have experienced more stress in storage at the source farm. 

Thus, as the season progressed and we could see the purchased stock wasn’t going to thrive, we had to pivot our plan to plant a mix of poplar and willow to just willow, but we expanded from the original 6 varieties (3 willow / 3 poplar) to include 9 varieties of willow we had also growing in the nursery, for other uses on the farm. We are working now with these varieties in our fodder plots. We don’t see this change as problematic since we are working to discover viable establishment strategies of fodder blocks and learning from the process along the way.  

Local cuttings were taken in the last two weeks of March, and stuck immediately into beds. We prepped in our field near the greenhouse, as well as in our air prune beds in the “woodland nursery.” We did this to provide a buffer of options in case something went wrong. We were thinking more in anticipation of possible drought, which might stress the field cuttings, but in fact the major event was a very warm April that led to a lot of quick leaf out of the cuttings, followed by a very devastating hard frost on May 17, where it was recorded to drop to 23F at our greenhouse weather station. The combination of rapid leaf out from the oddly warm spring and this event dealt a devastating blow to the cuttings in the field planting, yet the cuttings in the forest nursery were less harmed and recovered almost 100% across the planting. 

This spring, we also did site analysis and prep, sampling the three planting sites on 4/5/2023 and receiving results from the DairyOne lab for each site. All three sites have low Phosphorus (common on NY pasture lands) and high levels of K, Ca, and Mg. The pH for all sites was around 6.2 - 6.4 so suitable for willow, the most notable outlier was that each site had relatively high organic matter with site A at 8.2%, site B at 11.3%, and site C at 9.2%. 

Site prep included excavator work in April, to address water flow issues in the B to C area and mowing site A. Some adjustments to the original plan to differentiate site treatments was necessary due to some unexpected characteristics the sites presented. We were able to tarp Site A on August 7, which was sufficient after 3 months to remove and plant. We decided against tillage in any of the sites because the wetness of the soil didn’t allow for a BCS to work effectively. And instead of three distinct fencing systems, we decided to focus on a perimeter 3D fence that encapsulated the three sites, since we were also installing a 3D fence on the other end of the farm as part of another agroforestry tree project. 

A major factor in decision making this season was the substantial and persistent rain we experienced after a six week dry period in April and May. The cuttings looked well developed in Fall and so we prepared to plant out the sites. We were able to plant out Site A on November 10, but Sites B and C were heavily disturbed due to a new water line that was being installed. While we hoped the soil would dry out enough to plant, ultimately we decided it was best to wait to plant out the remaining sites in Spring 2024 and cover the remaining cuttings in the nursery with a deep bed of wood chips to protect from freezing temperatures. The 3D fence was fully completed and working with 8 - 10k volts by December 5, after much troubleshooting! 

Outputs

In addition to the field work described above, we are working on some materials that will be part of the final share out of the project. 

Initial database/literature review of Willow; we are working with Jeff Piestrak of FLX Agroforestry Solutions to review existing research and develop profiles of willow as a fodder species and and specific information on the cultivars we’ve settled on using

3D Fence Article: We have a draft of an article describing our process of planning and installing a 3D perimeter fence to reduce deer pressure on tree plantings. We’ve learned a lot or valuable lessons we are eager to share with others so they can save time and money if they choose to install one. Overall the fence is working well as a deterrent. 

Key Learnings

- The Austree and SX-61 hybrids far outgrew other cultivars in year one

- Goat willow seems like a promising and desirable fodder species but is also slow growing initially and needs extra tender care to thrive

- Sourcing material locally (including having a friend who had cultivars grown out) was much more cost effective than purchasing plants, especially when considering the high mortality of the purchased stock. In a few hours we had more material that we knew what to do with, and cuttings were easily less than 10 cents a piece, vs .50 - 2.90 purchased. The biggest advantage of locally sourcing is the control one has to the timing of the activity to keep cuttings fresh and viable. 

- Dormant hardwood cuttings are vulnerable to the variable spring thaw/freeze/warm weather conditions and that woodland nursery settings might buffer some of these effects and increase survival rates

- Overall, riparian areas are wonderful opportunities for developing fodder banks of wet tolerant species such as Willow. But, with the natural water dynamics of these zones, coupled with the increased variability from changing environmental conditions, means that development needs to be well timed, and even flexible, to plant out the space. Extra consideration to designing for water quality is foremost in importance, with the trees being a helpful follow up activity after the hydrology is addressed. Site B will be of particular interest, since we took advantage of the water line excavation and dug out substantial catches to slow water down and create berms for establishing the willow come Spring, 2024.

2024 Update

This season we have been able to successfully establish all three sites with randomized and rooted willow cuttings. We are happy to report this milestone to have the trees in the ground and growing.

Additionally, we acquired three additional species of larger diameter stakes from our technical advisor and decided to advantageously plant these along the edges but outside the “official” zone of the research. These varieties were recommended as being robust, producing ample vegetation, and growing quickly. One clear distinction is that we planted these in clusters of about a dozen “live stakes” spaced about 3 feet apart, using material that was around 1” in diameter. Given the wet winter that continued into spring, these stakes initially did very well, as did the plantings we established in Sites 2 & 3. 

Additional “clustered” willow added to the sites in 2024

While we had planted Site 1 in the previous Fall after removing the tarps, as noted last year the conditions were not favorable for planting the other two sites, because of excessive rain, the need for the excavated site to “rest” and settle, and the onset of colder than expected temps in November 2023. This provided a good opportunity to assess our progress and plan for a spring establishment of Sites B and C, respectively. This also gave us a second winter to test and assess the efficacy of the 3D fence before planting out more trees. 

Further, we continued to reduce the variability of the plots, since it proved difficult to see how we would effectively mulch and maintain plantings with the items originally listed above (sawdust / wood chips / weed whack). It's safe to say that a huge learning from this project is the necessity to streamline activities for site prep, protection, and maintenance of plantings. We’ve landed on a helpful strategy to get trees established and maintain them cost-effectively, as follows: 

1) Conduct site prep one season in advance, no matter the method.

2) Establish a 3D fence in advance of plantings, to encourage deer to re-route (more on this below) before establishing plantings.

3) Plant stakes that are ¾” or less in air prune beds for 1 - 2 seasons before planting. Stakes that are 1” or greater can be pointed at the bottom and driven in with a rubber mallet directly.

4) Immediately after planting, install a short tree tube and fiberglass stake to protect the base of the tree from burrowing rodents, and weed whacking. A taller stake also helps identify where the tree is if vegetation gets out of hand.

5) Weed whack strategically with timing, first at the end of the spring flush of growth (late May into early June) and then in late July/early August before anything gets too woody. Doing this at the right time drastically reduces the time it takes to complete the task. 

The summary of plots has been updated below to reflect these learnings. Site C should be noted as having an additional activity that was unplanned when this project was conceived; we had a forestry mulcher onsite in the spring and decided to reduce much of the overgrown, dying, “invasive” buckthorn that was growing in the overstory, since it was easier to manage before planting. This provided an interesting and dense layer of shredded trees that acted much like a thick mulch, but proved more difficult to plant into so soon after. If we could have waited a year it would have been more amenable, likely. 

Updated table with treatments to the three plots

Notes on the Deer Fence

While we are working on a longer article describing the installation and maintenance of a 3D style deer fence as part of the project outputs, in the meantime some important notes to share in reporting include: 

1) Changing patterns of behavior: It has become clear through the project that a 3D fence is a deterrent, not a fortress that will keep deer out perfectly. It has become necessary, then, to keep apprised about the actual specific deer that are interacting with a site and know their habits may change over time. In consulting for the project, we leaned on our neighbor who hunts our land to help describe the normal patterns of movement, noted on the map below (ORANGE). Rather than attempt to keep the deer out, what our plan of the fence aimed to do was encourage the deer to go around the planting zone, in order words stick to the outskirts and perimeter rather than cut through our field. 

2) Given the changing nature of individuals and herds, along with the effect of seasonality, it became necessary to therefore conduct routine inspections of the fence for evidence of deer, as well as testing and keeping the voltage up. Fallen limbs, uprooted stakes, and lines grounding out on saplings were all discovered during (almost) weekly check ups we conducted from May - October. We treated these as a chance to walk the land, in most cases taking less than one hour to complete. 

3) When deer pressure was observed, action was taken to address it. In one instance, a corner of the fence (marked with a yellow arrow on the map) became an easy access for deer since they could get a running start and hop the fence. We addressed this by adding an additional loop of fence to the system (also in yellow) in October 2024 as well as felling a bunch of trees to discourage a good landing spot beyond the perimeter fence. This seems to have helped as the evidence of deer in the site was eliminated in the following weeks. 

Still, even with these efforts we had some minor damage to willows planted, mostly the clustered “extra’ plantings we did. Yet since the deer may feel compelled to keep moving through the space, the damage was minor and we expect the trees to recover in the coming season. 

All this to say, between the need to check on the fence to ensure it is operating at its top potential, along with the timing of vegetation growth to hit it before it become unwieldy, routine visits to the site and fence became necessary, and will continue to be so, until the majority of the willow is above browse height. Time will tell if this will happen in 2025, or beyond.

Notes on Measurements

In addition to the establishment goals for the project we set out to take measurement of growth, which we have done for trees in the nursery beds and after planting. Initially in the proposal we suggested monthly measurements but this proved too time consuming, and not particularly useful. So, we are shifting to simply annual measurements of growth (2023); from cutting to first year, and then for two seasons in the planting. (2024 and 2025). This should offer an average amongst the various cultivars that will provide ourselves and other farmers with good data to make future decisions. 

Additionally, we have tracked the time and other costs associated with establishing willow in each of the three sites, which may prove useful when comparing the success of the different plantings. While the data won’t be completed and shared until our final report at the end of 2025, we can offer some initial observations of the plantings:

Generally, the trees planted in the excavated site (B) are growing faster and healthier than the other two sites. There is notable mortality and the trees are struggling most at site A, which may be a function of site prep (tarping), the timing of planting (Fall 2023), or some combination of these factors. The trees at site C seem to be doing initially pretty well.

We look forward to comparing the soils data, tree growth data, and observations from 2025 to see how these three sites compare at the end of the project.

Outputs

In addition to the field work described above, the following have been developed in 2024:

We built upon the initial database and literature review of willow and began outlining a guide to planting willow for Agroforestry, including information on collecting and care for cuttings, cultivar selection, site prep, and planting learned from this project. 

We still have an evolving draft of the 3D fence article to finalize in the coming year and share as well towards the completion of the project. 

Key Learnings

- Streamline the system: tree and fence maintenance is easy to fall down to the bottom of the list when so many other demands persist on the farm. It is necessary to find ways for these tasks to be achievable and routine. 

- Size of willow matters: during propagation, staking material should be sorted for nursery growing versus direct planting into the field. Once growing, the race to get above browse height and become established helps reduce ongoing maintenance demands. 

- Site prep is best a season ahead; When possible, the disturbance to a site takes time to occur, and time to recover from. Ideally a space of 6 - 12 months from site prep to planting seems best, at least within the context of the methods we have utilized so far.

- Deer fence needs vigilance: The 3D fence is a deterrent only if it's kept up, high voltage, and assessed for possible vulnerabilities on a routine (weekly?) basis. 

- Excavation is doing best, overall; perhaps not surprising, the excavation site is proving to provide a really nice growing medium for the trees. It's not a stretch to imagine that hilling up two to four feet of rich soil would result in such positive outcomes. 

2025 Results and Analysis

Growth in the three plots continued and we’re excited to share a collection of data, learnings, and reflections from this year and the project as a whole. In total, 79 trees were planted and 67 of them were still living by year 3. We were pleased to see an overall survival rate of 85% but also believe that the story goes beyond just “survivability” and that it is also important to look for “thrivability.” 

Regarding thrivability, we are looking at how quickly trees get established and grow above herbivore browse height. To measure this, we analyzed the year 3 growth data against thresholds of 48 inches and 72 inches. While we found that measuring against 48 inches is a good general baseline, it still leaves trees vulnerable to herbivore browse. Because of this, we also analyzed the percentage of trees within each plot and species that grew to above 72 inches which is a much safer threshold when a farmer can consider removing tree fencing and protection.

We’ll begin with updates on our data collection and analysis and a selection of charts and graphs. To view the complete set of summary tables and line graphs, you can view our Tables and Charts document as an attachment to this report. 

Below is our first summary table (Figure 1.1) from our final year 3 measurements. It summarizes the data across the 3 different test plots. 

Analysis By Plots

As shown in the data from Figure 1.1, Plot B (Excavated and Mounded) supported the most growth with an average tree height of 114.8 inches. We believe that this can possibly be attributed to the large disturbance event during the excavation in the plot which helped limit competitive vegetation throughout the establishment period. Plot B also had consistent moisture levels which we believe also supported this site’s strong growth and establishment. Of the 27 trees that were planted in Plot B, only three of them had died by year 3. If we remove the three deaths from the data, the average height of the living trees in Plot B increased  to 129.2 inches. 

Plot B also featured the most amount of trees that grew to above 72 inches. As you’ll see in Figure 1.1, 77.78% of the trees planted in Plot B grew to above 72 inches by year 3. This suggests that if you have the ability to do more site preparation up front by excavating and mounding planting locations, then you may be able to remove deer fencing and overall protection earlier. 

The plot with the next highest growth was Plot C (Swamp with topsoil spread) with an average tree height of 83.7 inches. This site also had consistent moisture as well as limited vegetative competition due to a large decomposing tree that was covering most of the soil. Out of the 27 trees that were planted, 6 of them died by year 3. This plot featured the lowest survival rate of 78% when compared to the survival rates of Plot A at 88% and Plot B at 89%. That said, if we remove the 6 deaths from the data, the average height of the living trees in Plot C increased to 105.2 inches. Because there was a lot of tree debris in this site, there was very little competitive vegetation to begin with and after planting we noticed that there still wasn’t a lot of understory growth. This condition, which was similar to the limited competitive vegetation in Plot B, may have contributed to the high growth rates in these top performing plots.  

The plot with the lowest growth was Plot A (Tarped and woodchipped) with an average tree height of 65.1 inches. 25 trees were planted and by year three, 3 of them died. If we remove the 3 deaths from the data, the average height of the living trees in Plot A increased to 74 inches. 

In Plot A, we also observed a noticeable difference in the percentage of trees that grew to above 48 inches but not 72 inches. This suggests that 48 inches was a meaningful threshold to measure growth against. 

Initial Growth as an Indicator

In Plot B, we noticed there was the largest growth from year 1 to year 2 when compared to the other plots. Considering that Plot B was also the tallest growing plot, this may suggest that early fast growth from year 1 to year 2 could be a good indicator of the speed it would take for a tree to grow above browse height. This means that one could also infer that if a tree’s growth is not substantial from year 1 to year 2 in the field, then it will probably take several years for it to grow above a browse height of 72 inches. 

Similarly, in all plots, and especially Plot C, if trees grew faster from year 1 to 2, it was typically a good indicator that strong growth would continue from year 2 to year 3

Now let’s move to Figure 1.2 and discuss the general findings across all of the 9 varieties of willow. 

On Average Height by Variety

Among all the 9 varieties, we found that they naturally fell into three groups: Top performers (4 varieties), middle performers (3 varieties), and low performers (2 varieties)

The 4 varieties that grew to the tallest average height were Flame, Schwernii, Sx61, and Austree. The middle performers were Weeping, Osier, and Rubykins. The low performers included Curly and Goat. 

Flame and Schwernii as Top Performers

Flame, with 9 trees planted, suffered zero deaths and had the highest average growth by year 3 at 136.2 inches. It displayed particularly spectacular growth in Plots B and C with average growth heights of 181.7 inches and 157.3 inches, respectively. 

Below you will find four line charts for the top 4 tallest growing varieties. Across all of these line charts, you will see a black dotted line. This line indicates the average trend across this variety. 

In general, across most varieties, we found that the orange line, representing Plot B, was typically above the average trend line and that the yellow line, representing Plot A, was typically below the average trend line. While this is not true for every variety, we did notice it as a general pattern. 

Following closely behind Flame was Schwernii, with 7 trees planted and zero deaths by year 3. Schwernii also had remarkable growth with an average growth by year 3 at 132.4 inches. In Plot B, Schwernii’s growth was particularly extraordinary with the two trees planted in Plot B growing to 183 inches and 206 inches! 

Sx61 (8 planted) and Austree (6 planted) also had noticeably strong growths with only 1 death each and year 3 average heights of 115.5 inches and 106.5 inches, respectively. If we remove each of the deaths for these varieties, the averages bump up to 132 inches for Sx61 and 127.8 inches for Austree. 

The next 3 performing varieties, with respect to average growth in height by year 3, were Weeping, Osier, and Rubykins with average heights of 105.3 inches, 78.4 inches, and 75.1 inches respectively. These average heights, like the average heights listed above for Schwernii and Sx61, included any deaths that occurred by year 3. If we exclude the deaths from the average height data for Weeping, Osier, and Rubykins, the averages increased to 118.5, 85.5, and 120.2 inches, respectively. 

In regards to this middle performing group, it is noteworthy to share a bit more about the results from the Rubykins variety. 8 Rubykins were planted and 3 of them died. Of the 5 that survived, their growth measurements were fairly distributed with year 3 heights of 40, 69, 126, 172, and 194 inches. This amounted to an average of 120.2 inches and the updated line chart shows this in Figure 9.3. Based on this small sample size, one could hypothesize that when Rubykins is happy with its growing conditions and environment, it has the potential to grow quickly. We noticed that it grew particularly well in Plot B which featured the mounded site preparation. That said, we also observed that Rubykins did not produce the largest or most dense leaf fodder, showing that even if a variety grows tall and fast, it isn’t necessarily the best choice for a tree-fodder application.

Finally, the lowest 2 performing varieties, with respect to average growth in height by year 3, were Curly and Goat with average heights of 52.9 inches and 28.7 inches, respectively. For both of these “low performers”, 9 varieties were planted and 2 died in each group. For the Goat variety, we measured three of the trees having lower heights measured in year 2 versus year 1. One of them died but two of them bounced back and continued putting on growth in year 3. 

On Browse Height by Variety

One of the main goals of this research was to see which varieties, if any, would grow to above deer browse height by year 3. We used two height thresholds, 48 inches and 72 inches, to measure our data against. Of the 9 varieties, 8 of them had at least half of their trees grow to above 48 inches. Despite this promising result, it is important to note that deer browse would likely still persist at heights around 48 inches. When we looked at the higher threshold of 72 inches, we found that all of the varieties, except Goat, had at least 3 trees grow to above 72 inches and 5 of the varieties had at least 70% of trees grow to above 72 inches. 

To get more specific, let’s go a bit deeper on a few of the varieties. For Flame, Schwernii, Sx61, and Austree, at least 78% of the trees made it above the 72 inch threshold by year 3. Among the 30 trees planted in these four top performers, 83% of them grew to above 72 inches by year 3. 

For Sx61, we noticed solid height growth across all three plots! If we exclude the one death from Plot A, we found that the average height of its growth across Plot A, Plot B, and Plot C was 132.5, 129, and 136 inches, respectively. This may suggest it has the ability to thrive in a variety of settings and environments.

For Weeping, all of the surviving trees made it above 48 inches with all but one making it above 72 inches. 

For Osier, the growth data was solid, but not excellent. This was a bit surprising given how well we’ve seen it thrive over the last 10 years on our farm. 92% (11 out of 12 planted) survived, and 83% made it above 48 inches. With that said, of the 11 survivors, only 7 of them grew to above 72 inches by year 3. 

The lowest performing varieties were Curly and Goat. Of the 9 Curlys planted, 2 died, 5 grew to above 48 inches, and only 3 made it above 72 inches. For Goat, only 2 of the 9 trees planted grew to above the 48 inch threshold with 0 making it above 72 inches. Despite its low growth heights, we did observe that the leaf biomass on the Goat variety was denser than the other varieties. This suggests that Goat may still be a promising tree fodder source if the farmer has the time to wait for it to grow above browse height and the resources to protect from deer browse before then. 

Overall, we found that 48 inches and 72 inches were useful thresholds to measure up against. When analyzing the growth results up against these thresholds, we noticed that Curly, Osier, and Goat had a 22%+  dropoff on how many trees grew to above 72 inches versus 48 inches. 

Key Findings

• Plot B’s site preparation (excavating and mounding) produced the best results of any plot with the tallest average heights and highest percentage of trees growing above 72 inches. This suggests that if you have the ability to do more site preparation up front by excavating and mounding planting locations, then you may be able to remove deer fencing and overall protection earlier. The results from Plot B are perhaps not surprising given that we mounded up two to four feet of rich soil before planting 
• The plot with the lowest growth was Plot A (tarped and woodchipped) with an average height by year 3 of 65.1 inches and 74 inches when excluding the 3 deaths
• Plot C had the most deaths as 6 of the 27 trees planted did not survive to year 3. That said, the ones that did survive grew to an average height of 105.2 inches by year 3. This places Plot C as the middle performing plot in terms of average growth height by year 3. Plot C performed similarly to Plot A (47.62%) regarding percentage of trees that grew to above the 72 inch browse height threshold by year 3 with 51.85%
• The hybrid varieties were some of the fastest growers, especially Flame, Schwernii, and Sx61
• Flame, Schwernii, Sx61, Austree, Weeping, were the top performers regarding average height and % above a browse height of 72 inches. Osier was also a steady grower but only 58% of the trees grew to above 72 inches by year 3. 
• Flame, with 9 trees planted, suffered zero deaths and also had the highest average growth by year 3 at 136.2 inches
• Schwernii, with 7 trees planted, suffered zero deaths and had the second highest average growth by year 3 at 132.4 inches
• For Sx61, we noticed solid height growth across all three plots. This may suggest it has the ability to thrive in a variety of settings and environments
• Goat and Curly struggled to grow tall by year 3 but that doesn’t mean they are bad varieties for fodder production over a longer time period if the farmer can protect them until they grow to above browse height
• For Goat, only 2 of the 9 trees planted grew to above the 48 inch threshold with 0 making it above 72 inches 
• Just because a tree grows fast, does necessarily mean it will be great for fodder, as it may have a low density and size of leaf material (i.e. Rubykins)

All tables and charts can be viewed here: Willow_-Tables-and-Charts