Decline of Casuarina equisetifolia: A Loss to Pacific Island Agroforestry

Decline of Casuarina equisetifolia: A Loss to Pacific Island Agroforestry

Summary

During year two of this grant, outreach activities were conducted to inform the local and scientific community of ironwood decline. A source of superior ironwood tree seeds from around the world was identified and will be grown and planted around Guam to increase the trees genetic diversity. An Ironwood Tree Committee consisting of producers and scientists from Guam, Australia, and mainland United States was established to analyze ironwood tree decline on Guam. Finally, statistical analysis of ironwood tree survey data determined that the presence of conks, termites and level of human intervention could explain the observed decline of Guam’s trees.

Objectives/Performance Targets

Objectives/performance targets for year two (September 1, 2009 to August 31, 2010) as listed in proposal:

Objective 4: To inform the public of the survey finding and to form an Ironwood Tree Decline Committee.

Objective 5: Identify a source of seeds from superior ironwood trees that the Guam Department of Agriculture can use in their give-away program.

Objective 6: Based an analysis of the data collected in Objectives 1-2 and 1-3, a conclusion will be drawn as to the cause or causes for ironwood decline.

Accomplishments/Milestones

1) A source of superior Casuarina eqisetifolia seeds from around the world has been identified and will be planted across the island of Guam to increase the genetic diversity and disease resistance of the ironwood tree on Guam. These seed lots were from international provenance trials and were collected by the Australian Tree Seed Center. The geographical locations of these seed sources are Australia, Thailand, China, Taiwan, India, Sri Lanka, Vietnam, Kenya, Honduras, Papua New Guinea, Pakistan, Egypt and Malawi.

2) An Ironwood Tree Decline Committee has been established:

Robert L. Schlub, PhD, Extension Specialist / Professor
University of Guam, ANR/CES/CNAS, UOG Station, Mangilao,
GU 96923; Phone 1-671-735-2089; Fax 1-671-734-5600;
rlschlub@uguam.uog.edu

Dilip Nandwani, PhD, EIPM Coordinator Saipan,Rota,and Tinian
CREES, Northern Marianas College, P.O. Box 501250, Saipan MP
96950; Phone 1-670-234-3690 ext 1725;
dilipnandwani@yahoo.com

Brian D. Marx, PhD, Professor, Department of Experimental
Statistics,Louisiana State University, Rm 141 Agriculture
Center, Agriculture Administration Bldg., Baton Rouge, LA
70803-5606,Phone: 1-225-578-8366; Fax: 1-225-578-8344;
bmarx@lsu.edu

M. Catherine Aime, PhD, Assistant Professor, Louisiana State
University,Agricultural Center, 455 Life Science Bldg.
Baton Rouge, LA 70803;Phone: 1-225-578-1383;
Fax: 1-225-578-1415; MAime@agcenter.lsu.edu

Alejandro Badilles, IPM Coordinator, Northern Marianas
College, CNMI, Rota, MP 96951; Phone 1-671-532-9513;
Fax 1-671-532-9512; abadilles@yahoo.com

Phil Cannon, PhD. Regional Forest Pathologist, USDA Forest
Service, 1323 Club Dr., Vallejo, CA 94592;
Phone 1-707-562-8913; Fax 1-707-562-9054, pcannon@fs.fed.us

Lisa Kennaway, Geographer, USDA-APHIS-PPQ-CPHST, 2301
Research Blvd. Suite 108, Fort Collins CO 80526-1825; Phone
1-970-490-4463; Fax 1-970-490-4479;
lisa.kennaway@aphis.usda.gov

Aubrey Moore, PhD. Extension Agent, ANR/CES/CNAS, UOG
Station, Mangilao, Gu 96923; Phone 1-671-735-2086;
Fax 1-671-734-5600; amoore@uguam.uog.edu

Scot Nelson, PhD, Specialist (Plant Pathology), University
of Hawaii at Manoa, CTAHR, PEPS, Komohana Research and
Extension Center, 875 Komohana St., Hilo, HI 96720;
Phone 1-808-969-8265; fax 1-808-981-5211; snelson@hawaii.edu

Melodi Putnam, Senior Instructor; Oregon State University,
Cordley Hall, Corvallis, OR 97331-2902;
phone 1-541-737-5542; putnamm@science.oregonstate.edu

Jason Smith, PhD, Assistant Professor, Forest Pathology,
University of Florida,134 Newins-Ziegler Hall, P.O. Box
110410, Gainesville, FL 32611-0410;
Phone 1-352-846-0843; fax 1-352-846-1277; jasons@ufl.edu

Pauline Spaine PhD, Biotechnologist, USDA/APHIS/BRS,
Plant Pest and Protectants Branch, Riverdale, Maryland 20737
Phone 301-734-0938, Fax 301-734-8669;
Pauline.C.Spaine@aphis.usda.gov

Bal Rao, Ph.D., The Davey Tree Expert Company, Ohio
Manager of Research and Development, 1500 N. Mantua St.,
Kent, OH 44240 Phone: 800-447-1667 ext. 351; brao@davey.com

Bernard Watson, Producer/Farmer, Guam
P.O. Box 20487, GMF, GU 96921
Phone: 671-687-2139

Felix Quan, Producer/Farmer
P.O. Box 12596, Tamuning, GU 96931
Phone: 671-637-7986

Juan Pangilinan, Producer/Farmer
P.O. Box 22335, GMF, GU 96921
Phone: 671-888-9546

Russell F. Young, Golf Course Superintendent
P.O. Box 315661, Tamuning, GU 96931
Phone: 671-366-7196, Fax: 671-362-7231, youngolf@guam.net

3) Further analysis of decline symptoms were conducted during this reporting period. Branchlet (“needle”) weight under different states of declining were determined (Fig. 1). The differences between the weight of branchlets were not significantly different between DS 0 (healthy) and 1 (slightly declined) or DS 2 and 3. However, DS 4 trees (severely declined) were the worst with 95.3% fewer branchlets when compared to DS 0 trees.

4) A second survey (Survey II) was conducted during the second year to reassess the decline severity from the year one survey (Survey I). Ironwood tree decline from survey II remained largely the same as survey I at nine sites, had increased at 17 sites and had decreased at 12 sites (Fig. 2). Alarmingly, IWTD is now appearing at previously healthy locations such as Cocos Island. During Survey II it was also determined that Guam’s ironwood population is comprised of 80% monoecious, 7% sterile, 13% dioecious trees of which 3% are males, and 7% were unknown (no flowers during survey).

5) Statistical analysis were performed on tree data collected from August 2009 to December 2009 (Fig. 3). For each sampled tree, the level of decline was measured on an ordinal scale consisting of five categories ranging from healthy (DS=0) to near dead (DS=4). Several predictors were also measured including tree diameter, fire damage, typhoon damage, presence or absence of termites, presence or absence of basidiocarps and various geographical or cultural factors. The five decline response levels can be viewed as categories of a multinomial distribution, where the multinomial probability profile depends on the levels of these various predictors. Such data structure is well-suited to a proportional odds model, thereby leading to odds ratios, involving cumulative probabilities which can be estimated and summarized using information from the predictor coefficient. Various modeling techniques were applied to address data set issues: reduced logistic models, spatial relationships of residuals using latitude and longitude coordinates and correlation structure induced by the fact that trees were sampled in clusters at various sites. Among our findings, factors related to ironwood decline include basidiocarps, termites and level of human management.

Response variable

Dec_sev (decline severity) is an ordered categorical variable from 0 to 4. If an observed tree is in perfect health, then it is recorded in the data as having dec_sev = 0 (the less healthy the tree, the higher the decline severity number.)

Dieback is a binary variable. If an observed tree was assigned a dec_sev greater or equal to 1, then the dieback value for that observation is equal to 1, otherwise the observation has a dieback value of 0.

Explanatory variables

Structure variables: num_stem represents the number of tree main stems. CBH stands for circumference at breast height, 1.5 m above ground. Density is defined as the number of ironwood trees per square meter at a surveyed site.

Stress variables

Fire, conk, typhoons and termites are all binary categorical variables. A tree displaying any signs of stress was assigned a 1. Trees with no stress were assigned a 0.

Geographic variables

Lat (latitude), long (longitude) and altitude were measured for each sampled tree using a GPS device. Site is an arbitrary number assigned for a location where the tree was sampled. If 30 sampled trees were located on one site, then all would be given the same site number. In total, there were 44 sites.

Miscellaneous variables:

Human_mgmt (0, 1, 2) assesses the level at which a sampled tree was managed [no management (0), intensive management (2)].

Planted_natural is a categorical variable and refers to tree establishment: planted(p) or natural (n).

STATISTICAL MODELS:

Logistic Regression

Logistic regression is a form of analysis in which a binary response of 0s and 1s is fitted by a set of continuous and/or categorical explanatory variables. The model determines the probability that an experimental unit belongs to a group. In the context of the ironwood tree data, the probability of a tree experiencing dieback is generally unknown but can be estimated by a set of explanatory variables. The basic form of a logistic regression is:

The formula on the right hand side resembles a standard linear regression setting, where alpha is an intercept term and beta is a vector of coefficients for a vector of explanatory variables x. The left side of the equation is a log odds ratio. An odds ratio is a ratio of the probability of a tree having a dieback value of 1 over the probability of the complement, a tree not experiencing dieback.

Interpreting individual explanatory variables, in the context of a logistic regression, is determining which variables have a significant effect on the odds ratio. If an individual parameter ßi is equal to zero, this implies that it has no effect on the odds ratio. Confidence intervals for each explanatory variable can be constructed accordingly:

Multinomial Regression

Although a logistic model is adequate for explaining differences between unhealthy vs. healthy trees, it is not adequate for evaluating several levels of tree health. Therefore, a multinomial model is needed, specifically one that uses a cumulative logit link function. A model which uses a cumulative logit function is required because the decline severity index has a natural ordering. The basic function resembles the logistic regression function, except that the odds ratio is replaced with a cumulative odds ratio. Interpretation of parameter estimates follows within the context of a cumulative odds ratio; and as in a logistic model, if a parameter estimate is equal to zero, it implies that the parameter has no effect on the cumulative odds ratio. Confidence intervals for parameter estimates are constructed similar to the parameters from the logistic regression.

RESULTS

After fitting a logistic model with all explanatory variables, except for the longitude and latitude variables (to account for spatial effects), and applying several model selection algorithms, the final logistic model was reduced down to six variables (Table 1). Pearson residuals were interpolated along a two-dimensional grid using the longitude and latitude coordinates. Since the interpolated residuals fall in the range of -1 to 2.5, it suggests that the logistic model has an adequate fit (Fig. 4). Interpolating the Pearson residuals for the cumulative logit model resulted in extreme values; thereby, suggesting that the cumulative model may not be adequate (Fig. 5).

CONCLUSIONS

The logistic model for the current data was the better model. From the results of the logistic model, the most significant variables that could explain the ironwood tree’s state of health are the presence of basidiocarps, a high level of human management, or the presence of termites. The cumulative model, however, cannot explain trends in the data adequately. A reason could be that assumptions on proportional odds were not met. Possible ways to improve the cumulative model include adding more explanatory variables or moving away from the cumulative model to a multinomial model where the decline severity index is not considered ordinal.

6) The following publications/posters were submitted and presented during year two of this grant. This included a thesis by Karl (K.A.) Schlub. Mr. Schlub was awarded his Masters Degree.

Mersha, Z. and Schlub, R.L. 2010. Casuarina Dieback on Guam, In Forest Health Highlights, Pacific Islands April 2010 p. 7.

Mersha, Z., Schlub, R. L., Spaine, P.O., Smith J.A., and Nelson, S.C. 2010. Visual and quantitative characterization of ironwood tree (Casuarina equisetifolia) decline on Guam. Poster in proceedings of 2010 APS annual meeting Charlotte, North Carolina: Phytopathology 100:S82

Mersha, Z., Schlub, R.L., Spaine, P., Smith, J., Nelson, S., Moore, A., McConnell, J., Pinyopusarerk, K., Nandwani, D., and Badilles, A., 2010. Fungal Associations and Factors in Casuarina equisetifolia decline. Poster in proceedings of 9 th International Mycological Congress Edinburgh, UK: IMC9: P3.268

Schlub, K.A. 2010. Investigating the Ironwood Tree (Casuarina
Equisetifolia) Decline on Guam Using Applied Multinomial Modeling.
M.Ap. Stat. thesis, Louisiana State University

Schlub, K.A., Marx, B.D., Mersha, Z., Schlub, R.L. 2010. Investigating the ironwood tree (Casuarina equisetifolia) decline on Guam using applied multinomial modeling. Poster in proceedings of 2010 APS annual meeting Charlotte, North Carolina: Phytopathology 100:S115

Schlub, R.L., Mersha, Z., Aime, C.M., Badilles, A., Cannon, P.G., Marx, P.G. , McConnell, J., Moore, A., Nandwani, D., Nelson,S.C., Pinyopusarerk, K., Schlub, K.A., Smith, J.A., and Spaine, P.O. 2010. Guam Ironwood (Casuarina equisetifolia) Tree Decline Conference and Follow-up. Proc. 4th International Casuarina Workshop, March 22-25, Haikou, China

• Accomplishments Tables Graphs and Figures

Impacts and Contributions/Outcomes

1) Outreach activities for the general public included presentations on the importance of the ironwood tree to the island’s ecology and agriculture, how to care for ironwood trees and how to identify and reduce the impact of IWTD. Activities included advisement of advanced high school biology students on Saipan on the tree species Casuarina equisetifolia (Ironwood tree) and interactive displays at the University of Guam’s Charter Day and at the Environmental Protection Agency’s Earth Day. Hundreds of students, teachers, farmers and members of the general public were informed of ironwood tree decline and ironwood tree care as a result of these activities. The number of landowners and managers trained to develop Stewardship Plans during these events were approximately 52. The number of direct contacts who increased awareness of benefits and opportunities during these events was approximately 840.

2) Outreach activities for the scientific community included posters presented at three international professional meetings (the 4th International Casuarina meeting in Haikou, China, the annual meeting of the American Phytopathological Society, Charlotte, North Carolina, and 9th International Mycological Congress Edinburgh, UK); a thesis from the Louisiana State University Experimental Statistics department by Karl A. Schlub “Investigating the Ironwood Tree (Casuarina equisetifolia) Decline on Guam Using Applied Multinomial Modeling”; a visit and review of our activities from the USDA Region 5 Forest Pathologist Dr. Phil Cannon; and a presentation on the Northern Marianas island of Saipan entitled “Studies on the Decline of the Tree Species Casuarina equisetifolia in the Marianas Archipelago” for the Asia Pacific Academy of Science, Education, and Environmental Management.

Collaborators: