Final report for GW18-014
Project Information
Koa (Acacia koa), a native Hawaiian hardwood tree, has undergone vast reductions in population over the last 200 years from grazing and human activities. Replanting of koa has gained momentum over the last two decades for both conservation and production reasons, since koa provides ecosystem services and the timber is exceptionally expensive. In conservation or silvicultural settings, a major obstacle to koa establishment is competition of juvenile trees with weeds. The management options for this include mowing, herbicide application, or combinations of both, which are costly and hazardous to the environment. Another option that has received zero investigation is grazing the competitive understory with ruminants, after creating a conditioned feed aversion to the palatable, nutritious koa foliage. I will study the efficacy of conditioning sheep to avoid feeding on koa using Lithium Chloride (LiCl), a nausea-inducing chemical. The conditioning process requires a controlled setting, therefore after conditioning animals in pens I will evaluate if the aversion persists in a planted stand of koa. The length of time an aversion lasts has large variability, therefore I will quantify the persistence of conditioning.
Research Questions:
- Can sheep be conditioned to avoid feeding on koa foliage in a controlled setting?
- If the aversion is successful in confinement, will it persist in the field?
- How long will the conditioned aversion last?
- What effects will grazing a koa silvopasture have on koa and nutrients?
If aversions to koa foliage can be developed in sheep, producers can diversify and increase income while at the same time reducing negative environmental impacts. I will communicate the project results in printed protocol pamphlets handed out at demonstration field days held on Haleakala Ranch, publications in scientific journals, and presentations at national meetings.
Our objective is to condition sheep to avoid feeding on koa foliage in a young koa stand with non-native grass understory. We hypothesize that grazed koa silvopastures will have enhanced koa growth due to suppression of competition from understory grasses and higher concentrations of plant available nutrients in soil.
Research
Experimental Design
This study consisted of three treatments arranged in a randomized complete block design. Blocking was based on the study site being located on a mildly sloping (2%) hill. Each treatment was replicated three times for a total of nine experimental units. Treatments consisted of grazing at a standard stocking rate (approximately 50% aboveground herbage removal), mowing and an unmowed, ungrazed control. The study implemented four treatment events over the course of one year, based on the timing of forage regrowth.
Soil organic carbon and organic nitrogen concentrations were measured at the beginning and end of the study. Three soil samples per experimental unit were collected with a soil corer to a depth of 15 cm. Samples were air-dried, ground finely with a mortar and pestle, and analyzed for organic carbon and nitrogen concentration via a Costech Elemetal Analyzer (Costech, Valencia, CA). Several responses variables were measured immediately before and after each of the four treatment events, including forage aboveground biomass, sward height, and botanical composition. Forage aboveground biomass was collected by randomly placing three rings (each with an area of 0.25 m2) inside the experimental units, and clipping all of the forage inside the rings to 5 cm height. Biomass samples were collected pre- and post-treatment application to determine vegetation removal. Vegetation was separated by species, dried in an oven, and weight recorded on a dry basis.
Aversion Training and Field Grazing
There were four grazing events throughout the yearlong study, occurring in June and October of 2018, and February and June of 2019. Sheep were conditioned to avoid koa foliage in the following way. A herd of 15 mature ewes with no previous history of koa exposure were maintained over the course of this study and used at every grazing event. Prior to a grazing event, ewes were fasted overnight, and then provided access to koa foliage that had been recently harvested from living trees. Access to koa was provided in a dirt-floor pen, and foliage was suspended in a feed bunker. Both the smaller, actual leaves, as well as the photosynthetic elongated stems, or phyllodes, of the koa were presented to sheep.
After adequate intake of koa foliage by an individual sheep (>10 bites), that sheep was restrained and dosed orally with lithium chloride at a rate of 250 mg / kg bodyweight. This was done to develop a negative aversion to feeding on koa based on negative feedback provided by the lithium chloride in the form of acute nausea.
After all sheep in the herd were treated in this way, they were turned back out to pasture and allowed to graze freely until the evening, when they were collected and fasted again overnight. The following day, aversion was checked by providing treated animals with additional access to koa foliage. If any animals fed at this point, they were re-treated with lithium chloride. After this check, animals were turned on to experimental units that were sized to be moderately grazed after one day of residency of the experimental herd.
Bite counts were recorded in an effort to monitor sheep grazing behavior among the various vegetation classes present within experimental units. Additionally, branches of koa trees within plots were marked and the number of leaves on each branch was recorded to track sheep consumption of koa.
Statistical analyses
Data was analyzed with Minitab (Minitab, 2010) using General Linear Model. Replication was designated a random variable and Treatment and Event were designated as fixed variables. Replication was nested in Differences were considered significant at the p < 0.05 level. Each response variable was analyzed by treatment, event, and their interaction to discern the driving forces of variability observed.
Aversion Effectiveness
At the first grazing event, after no animals fed on koa in the feeding bunker on the second test day, the aversion persisted for the herd in the field for one day, but on the second day of grazing one animal selected for koa and another barked trees readily. These animals were removed before grazing the third replicate on the third day, and no koa consumption was recorded for the third day.
For the second grazing event, the conditioned herd was down to 10 due to one independent mortality in the field. Of the remaining 10, eight required re-aversion on the first test day, then none on the second test day. No koa consumption was recorded or observed over the three days of grazing the second event.
For the third grazing event, 10 sheep were still present, three necessitated re-aversion on the first day of testing, and none on the second. No koa consumption was observed on the first or second days of grazing, but a mild take was recorded on the third.
For the fourth grazing event, none of the 10 animals fed on koa provided in the bunker, thus none were treated with lithium chloride. Upon grazing of this round, the first grazed plot experienced no koa feeding, but the second and third plots had large amounts of koa foliage fed upon by at least seven individuals in the herd. Feeding on koa by animals who had previously avoided it occurred rapidly after those animals observed another animal feeding on koa foliage. This phenomenon is well documented in the literature (Ralphs, 1997) and accentuates the need for expedited removal of untrained animals.
That animals’ intake of feed in the bunker decreased throughout the grazing events, and consumption of koa in the field increased, indicates a potential problem with not implementing the conditioning in situ. Likely, the feed being offered in the bunker within the dirt-floor pen was not associated with the actual trees found in the field due to differences in the architecture of a live tree or cut limbs in a feed bunker. The need for conditioning to take place in the area where it is to be utilized is further noted by the high frequency of barking that took place within experimental units, which could not be averted against in a feed bunker trial. The aversion would likely have been more effective if it had been instilled in the field, where all components of the koa tree (i.e. bark) could be associated with negative feedback.
Sheep Grazing Behavior
On average, sheep bites were from forage 80% of the time, gorse 16%, koa 2%, and fireweed 2%. A spike in percentage of bites for gorse occurred in Event 2, when available forage was low from being out of season (Table 1). Another increase was observed in Event 4. In the latter case, forage was more available but the ever-increasing amount of gorse present within grazed plots limited selection for other vegetation classes.
|
Table 1. Bite count percentages from different vegetation classes of sheep feeding in koa silvopastures. |
|||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
|
|
|
-------------------------------------------%-------------------------------------------- |
||||
|
Forage |
89 ± 10 |
69 ± 17 |
83 ± 13 |
78 ± 22 |
|
|
Gorse |
9 ± 9 |
28 ± 19 |
10 ± 13 |
16 ± 25 |
|
|
Koa |
0 |
0 |
5 ± 9 |
5 ± 7 |
|
|
Fireweed |
2 ± 4 |
3 ± 3 |
2 ± 4 |
3 ± 2 |
|
Similar results were found when observing the percent of leaves consumed from marked branches in grazed plots (Table 2). The first event showed 18% consumption of marked branches, then zero for the second event, which coincided with no bite counts recorded for koa in Event 2. Increased removal of leaves in the latter two events was a result of the aversion breaking down progressively. The highest percentage of leaf removal occurred in Event 4 when no animals had been re-averted in the pen. This is evidence of conditioning against koa foliage being effective when immediate feedback was applied, but not carrying over to prolonged field conditions.
|
Table 2. Leaf removal percentage of marked branches, averaged over all branches and replications per event |
|||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
|
|
|
-------------------------------------------%-------------------------------------------- |
||||
|
Leaf consumed |
18 |
0 |
15 |
20 |
|
Canopy Characteristics: Sward height
Mowing was the most effective treatment for reducing sward height, both for each treatment event and over the entire course of the study (Table 3). Within each event, this was due to the homogeneous, close to soil level herbage removal via mechanical cutting.
|
Table 3. Sward height pre- and post-treatment imposition across four treatment events. |
|||||||||||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
|||||||||
|
|
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
|
|
|
-------------------------------------------cm-------------------------------------------- |
||||||||||||
|
Control |
74 |
70 |
-4 |
54 |
55 |
1 |
57 |
56 |
-1 |
90 |
88 |
-2 |
|
|
Grazed |
74 |
26 |
-48 |
51 |
27 |
-24 |
56 |
33 |
-23 |
76 |
42 |
-34 |
|
|
Mowed |
79 |
11 |
-68 |
34 |
10 |
-24 |
51 |
9 |
-42 |
77 |
9 |
-69 |
|
The control treatment had no effect on sward height within each event, but sward height increased over the course of the yearlong study under the control treatment. This was expected since the control treatment was essentially a fallow management practice where present vegetation is left free to grow.
Grazing had an intermediate effect on reduction of sward height between mowing and control treatments, both within each event and over the course of the study. This was mainly due to the prevalence of gorse within experimental units. Gorse was selected readily by feeding sheep, but the shrubby architecture of gorse coupled with prolific spines throughout the stems prevented feeding access to sheep for large portions of each plant. Therefore, although the forage component of the sward was grazed to much lower heights than those reported in the Post measurements of Table 3, the inability of sheep to fully graze gorse prevented them from lowering the overall sward canopy height.
Herbage Mass
The mowed treatment resulted in the highest total herbage mass removal (Table 4), due to high removal rates of both the forage (Table 5) and gorse (Table 6) components of the sward. This effect of higher total herbage mass removal for the mowed treatment was most pronounced in Event 1, and lessened throughout the proceeding Events, due much in part to the sustained reduction of gorse in mowed plots (Table 6).
|
Table 4. Total Herbage mass pre- and post-treatment imposition across four treatment events. |
||||||||||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
||||||||
|
|
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
|
|
------------------------------------------- g 0.25 m2 -1-------------------------------------------- |
|||||||||||
|
Control |
115 |
107 |
-8 |
118 |
109 |
-9 |
137 |
140 |
3 |
131 |
135 |
4 |
|
Grazed |
116 |
60 |
-57 |
115 |
75 |
-40 |
134 |
92 |
-43 |
129 |
91 |
-38 |
|
Mowed |
133 |
33 |
-101 |
44 |
15 |
-28 |
71 |
22 |
-49 |
64 |
19 |
-46 |
|
Table 5. Forage herbage mass pre- and post-treatment imposition across four treatment events. |
||||||||||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
||||||||
|
|
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
|
|
------------------------------------------- g 0.25 m2 -1-------------------------------------------- |
|||||||||||
|
Control |
99 |
93 |
-6 |
75 |
61 |
-13 |
84 |
86 |
1 |
62 |
64 |
2 |
|
Grazed |
89 |
28 |
-61 |
56 |
18 |
-38 |
64 |
40 |
-24 |
63 |
28 |
-35 |
|
Mowed |
110 |
28 |
-82 |
35 |
15 |
-21 |
69 |
19 |
-49 |
60 |
18 |
-42 |
|
Table 6. Gorse Herbage mass pre- and post-treatment imposition across four treatment events. |
||||||||||||
|
|
Event 1 |
Event 2 |
Event 3 |
Event 4 |
||||||||
|
|
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
Pre |
Post |
∆ |
|
|
------------------------------------------- g 0.25 m2 -1-------------------------------------------- |
|||||||||||
|
Control |
16 |
14 |
-2 |
43 |
47 |
5 |
53 |
55 |
2 |
69 |
72 |
3 |
|
Grazed |
27 |
31 |
4 |
59 |
57 |
-2 |
71 |
52 |
-19 |
66 |
63 |
-3 |
|
Mowed |
24 |
5 |
-19 |
9 |
1 |
-8 |
2 |
2 |
1 |
4 |
1 |
-4 |
Grazed treatments resulted in intermediate total herbage mass removal compared with mowing and fallow. Total herbage mass removal was less in grazed plots than in mowed treatments because of the inability of sheep to fully consume gorse. This is evidenced by the rising mass of gorse material throughout the extent of the study in grazed plots. As gorse plant became more mature within experimental units, sheep were less able to fully graze them due to excessive lignification associated with plant maturation.
Koa Growth
No treatment effects were significant on either koa height or diameter at basal height (Table 7). Koa tree height increased 40.2 cm, and DBH increased 19.5 mm on average over the course of the 1-year study. This lack of difference is likely due to relatively short time (< 1 y) to observe treatment effects on growth characteristics of trees.
|
Table 7. Koa diameter at basal height (dbh) and height at the beginning and end of treatment imposition, and their differences. |
||||||
|
|
Koa dbh Pre |
Koa dbh Post |
Koa dbh Change |
Koa Height Pre |
Koa height Post |
Koa Height Change |
|
Treatment |
---------------- mm ------------------ |
----------------------- cm -------------------- |
||||
|
|
|
|
|
|
|
|
|
Control |
40.4 (5.9) |
58.0 (6.6) |
17.5 (3.3) |
194.5 (26.0) |
238.3 (23.2) |
43.8 (5.1) |
|
Mowed |
43.2 (9.3) |
63.4 (13.5) |
20.2 (5.8) |
192.1 (33.7) |
226.8 (40.7) |
34.7 (11.9) |
|
Grazed |
47.0 (2.0) |
67.8 (7.7) |
20.8 (6.2) |
218.3 (17.4) |
260.4 (11.2) |
42.1 (8.1) |
Soil Characteristics
Treatment did not significantly affect either soil organic carbon or nitrogen concentrations (Table 8). There was a slight trend of the mowed treatment having a higher carbon and nitrogen concentration gain over the control and grazed treatments, but this was not statistically significant. This is probably due to the relatively short overall length of the study, and the long periods of time required for shifts in carbon concentration to be observable.
|
Table 8. Soil carbon and nitrogen concentration to 15 cm depth at the beginning and end of treatment imposition, and their differences. |
||||||
|
|
Carbon Pre |
Carbon Post |
Carbon Change |
Nitrogen Pre |
Nitrogen Post |
Nitrogen Change |
|
Treatment |
------------------------------------------ % -------------------------------------------- |
|||||
|
|
|
|
|
|
|
|
|
Control |
10.13 (1.36) |
9.95 (1.78) |
-0.17 (1.60) |
0.72 (0.06) |
0.69 (0.09) |
-0.03 (0.03) |
|
Mowed |
10.95 (0.47) |
12.69 (0.82) |
1.74 (1.00) |
0.78 (0.02) |
0.93 (0.06) |
0.14 (0.05) |
|
Grazed |
12.22 (2.52) |
11.95 (2.05) |
-0.27 (2.92) |
0.82 (0.13) |
0.88 (0.14) |
0.06 (0.12) |
Conclusions
Based on the results of this study, we found that sheep can be conditioned to avoid feeding on koa in a controlled setting, but that there are challenges in translating conditioning from a pen to the field. Due to this challenge, we were unable to determine how long an effective aversion can potentially last in a field setting. Further, this study supports the claim that animals that were not averted successfully must be removed from the herd, or will cause accentuated conditioning breakdown among the group. The producers we worked with on this project have incorporated these findings into their koa management plan, and will be carrying out future endeavors within the context of these findings. Mainly, they will avert animals to koa in the field, where all plant parts in their natural architecture can be conditioned against, and after conditioning has taken place, they will remove animals immediately that lapse and feed on koa.
Citations
Ralphs, M.H. 1997. Persistence of aversions to larkspur in naive and native cattle. Journal of Range Management 50: 367-370.
Research Outcomes
Education and Outreach
Participation Summary:
No outreach activities have taken place because of scheduling challenges. A journal article is in progress that will outline the findings of the study. The findings of the research were shared via personal communication to many members of the Hawaii Cattlemen's Council at their annual meeting in 2019.