Abstract. Six herbicides were applied at 3. 9, and 27 lb/A to the soil in forests of three types in Puerto Rico. The forests differed in soil type, amount and pattern of rainfall, and species composition. Defoliation occurred slowly, but the correlation between per centage defoliation and plants killed was significant at the 1% level of probability. Some defoliation was evident 1 month after treatment and increased slowly during the next 6 to 8 months. The most effective herbicide was 4-amino 3.5.6-trichloropicolinic acid (piclorani). Generally, 2-methoxy-3,6-dichlorobenzoic acid (dicamba), 2 methoxy-4.6-bis(isopropylamino)-s-triazine (prometone), and 5-bromo-3-1ec-buts 1-6-methyluracil (bromacil) were less effective than picloram. Both 2.3.6-trichlorophenylacetic acid (fenac) and 33.4-dichlorophenyl)-1,1-dimethylurea (diuron) were ineffective. Defoliation was greatest and more plants were killed in the driest

site.

One year after treatment, fenac was the most persistent herbicide in the soil. Herbicides were found to a 36 to 48-in depth within 3 months after application.

Herbaceous and woody seedlings were present at all sites within 18 months after herbicide application. There was no definite relationship between herbicidal treatment and secondary succession, except that the number and frequency of successional species were greater on plots having the highest percentage of defoliation.

INTRODUCTION

Tof woody has not studied intensively or extensively. A short review by Mayo-Menendez (7) indicated that frill treatment to trees was the most effective method of application. Beveridge (2), Sposta (11), and Wyatt-Smith (17, 18) used frill treatments successfully on a large number of tropical species but noted differential susceptibility among species. Dawkins (4) reported that a basal spraying or painting was as effective as frill treatment.

effect of herbicides applied to the soil for control

Many hardwoods can be controlled effectively with phenoxyacetic or picolinic acids applied in basal frills or as tree injections (6, 15). Many compounds are effective as foliar treatments for controlling woody plants (3, 16). Nation (9) reported that 4-amino-3,5,6-trichloropicolinic acid (picloram), when applied as a soil treatment, was an effective herbicide for controlling woody plants. He also noted differential susceptibility among species. Inseparable from the biological effect of herbicides applied to the soil are their movement and persistence.

Received for publication February 15, 1967.

Research Agronomist. Agricultural Research Technician, Research Range Scientist. respectively, Crops Research Division. Agricultural Research Service. U.S. Department of Agriculture, Federal Experiment Station. Mayaguez, Puerto Rico.

Factors such as rainfall, physical and chemical characteristics of the soil, microorganisms, chemical characteristics of the herbicides, and method of application may influence herbicidal movement and persistence (5, 8, 10, 12).

Our objectives were (a) to determine the toxicity of selected herbicides on tropical arboreal vegetation, (b) to determine the movement and persistence of these chemicals in the soil, and (c) to observe the short-term effects of these chemicals on secondary plant succession.

MATERIALS AND METHODS

Our studies were located in Guanica Commonwealth Forest, Maricao Commonwealth Forest, and Luquillo National Forest. The soil type at the Guanica Commonwealth Forest is Jacana clay. It is an alluvial soil normally less than 36 in deep, with very low permeability. The vegetation is xerophytic. There were 35 woody species within the test area, but Leucaena leucocephala (Lam.) DeWit and Haematoxylon campechianum L. comprised more than 90% of the woody plant population. Mean height of the vegetation was 15 ft. Annual rainfall in the Guanica area, estimated at approximately 30 in, occurs largely from July to October. The recorded annua' rainfall at the site for 1964 and 1965 was 27.88 and 25.00 in, respectively.

The soil type at the Maricao Commonwealth Forest site is Nipe clay, a permeable, well-drained, lateritic soil derived from serpentine. The vegetation on this site is classified by Beard (1) as moist tropical forest. There were 106 woody species within the test area. The mean canopy level was approximately 50 ft. Mean annual rainfall at the Maricao site is estimated to be about 90 in. Rainfall is normally distributed throughout the year, but December to May is the driest period. The recorded annual rainfall at the site for 1964 and 1965 was 84.64 and 109.88 in, respectively.

The soil type at the Luquillo National Forest site is Los Guineos clay loam, a plastic clay with poor internal drainage. The vegetation is a tropical rain forest (1) with a mean canopy level of about 60 ft. There were 88 tree species on the test site. Mean annual rainfall is estimated to be over 100 in. The highest rainfall normally occurs from July to October, but droughts are unknown. The recorded annual rainfall near the site was 85.78 in for

PICLORAM 9 LB/A

GUANCA MARKAO

69 12 15 18 MONTHS FOLLOW NC APPLICATION

Figure 1. Percentage of defoliation at specified intervals after picloram was applied to the soil in three forested areas in Puerto Rico.

remained essentially constant for the rest of the 24-month period.

Total vegetation control was short-lived, even on plots that had been treated with 27 lb A. No treatment prevented plant succession for more than 18 months. Two years after treatment, Ipomoea spp., Indigofera suffruticosa Mill, and cacti (Opuntia spp.) were abundant. Grass and tree seedlings were less numerous. Leucaena leucocephala (Lam.) DeWit was the most abundant tree seedling. Seedling density was directly related to per centage of defoliation.

Maricao. Picloram caused the highest percentage plot defoliation (Table 1). Dicamba at 27 lb A caused 50% defoliation, but this was considerably lower than that caused by picloram at 9 lb A. Bromacil, prometone, diuron, and fenac were considerably less effective.

Picloram affected the broadest spectrum of tree species (Table 3). Miconia sintenisii Cogn, was more susceptible to picloram than were the other eight species. Repence ferruginea (R. & P.) Mez appeared to be most susceptible to bromacil. Ocotea leucoxylon (Sw.) Mez tolerated all the herbicides except picloram at 9 and 27 lb ́A.

Maximum defoliation resulting from treatment with picloram was obtained at Maricao about 1 year after treatment (Figure 1). There was essentially no change during the second year of observation

No herbicidal treatment prevented the succession of tree seedlings, grasses and sedges, or vines and herbaceous plants at Maricao. Tree seedlings were sparse on plots treated with picloram at 9 and 27 lb A, but composition

Check

and density of the seedling complex could not be definitely associated with herbicidal treatment. The most common succession species in the Maricao site were Terebraria resinosa (Vahl, Sprague, Clusia spp, Ich nanthus pallens (Sw.) Munro, Panicum glutinosum Sw., and Polypodium spp.

Luquillo. Picloram caused the highest percentage of defoliation at the Luquillo site Table 1 The highest rates of bromacil and dicamba caused appreciable defoliation, but both herbicides were much les effective when only 3 or 9 lb A were used. Prometone diuron, and fenac caused only slight defoliation at all rates. Picloram was as effective at 9 ib. A as dicamba and bromacil at 27 lb 'A Differential susceptibility was apparent among the eight most numerous species Table 4. Tabebuia hetero phys DC Britton was the most resistant species in that no treatment caused more than 31% defoliation Prycho tria berterians DC on the other hand, was completely de

The speed at which trees defoliated was essentially the same for all herbicides applied. Data for the rate of defoliation caused by picloram indicate that maximum effectiveness was obtained about 6 to 8 months after treatment Figure 1). The maximum defoliation occurred sooner from picloram at 9 lb A than picloram at 27 lb 'A. Defoliation remained essentially constant for the remainder of the 24-month period.

The relative density of new seedlings at the Luquillo site could not be correlated with the herbicidal treatment. The most common tree seedlings were Psychotria besteriana DC... Ocotea leucoxylon (Sw.) Mez, and Picstoca montana (Graham) Nicholson, Grasses were best represented by Panicum adspersum Trin. and sedges by Scleria secans (L). Many vines and herbaceous plants were present. Ipomoea spp. being the most common There was a direct relation between percentage of defoliation and number of new seedlings. One year after treatment, the forest floor of plots that had been completely defoliated was covered with vegetation. The density of new plants is shown in Figure 2.

Herbicida ves the in the soil. Three months after application, the herbicides had moved downward in the soil to the 36 to 48-in depth. The bioassay data for all sampling depths indicated persistence of the herbicides in the soil 1 year after treatment for all locations was in the order of fenac > prometone > picloram > diuron > bromacil dicamba. An example of the residue data is shown in Figure 3. Dicamba had almost completely disappeared 1 year after treatment. Two years after treatment, fenac was still the most persistent herbicide. followed by prometone and picloram.

The persistence of the herbicides generally was greatest in the driest area Guanical and least in the wettest area Luquillo Figure 4 One year after application. the residue of picloram in plots treated at 27 lb A remained in relatively high concentrations at all test sites, as determined by the cucumber bioassay test. The presence of picioram in plots treated at 9 lb A could be eas detected 1 year after treatment, but the concentrations were about 10 times less than in plots treated with A The residue data for all locations indicated a trend for all the herbicides to dissipate more rapidly the top 12 in of soil

Over 200 woody species were represented on the three test sites, but several species were represented by only a few individuals. Tschirley (13) enumerated some of the problems involved in evaluating herbicides in tropical forests. Other workers (2, 9, 11, 16) have shown distinct differential susceptibility of woody plants to various herbicides. Differential species susceptibility also was evident at our test sites. When all the treatments at all locations are considered, Tabebuia heterophylla (DC.) Britton, Cordia borinquensis Urban, Inga fagifolia (L.) Willd., and Ocotea leucoxylon (Sw.) Mez were most resistant and Psychotria berteriana DC., Miconia sintenisii Cogn., Senegalia westiana (DC.) Britton & Rose, and Leucaena leucocephala (Lam.) DeWit were most susceptible. Cascaria-Drypetes, Miconia sintenisii Cogn. and Miconia prasina (Sw.) DC. were represented at both Maricao and Luquillo. The reaction of these species to the herbicides followed the same general trend at both locations, but defoliation was greater at Maricao.

The effect of climatic and edaphic factors on herbicidal activity cannot be clearly elucidated in this study because of differences in species composition. At all three test sites, rainfall before application was sufficient to maintain adequate plant growth. Rainfall for 2 months after treatment was 2.41 in at Guanica, 5.16 in at Maricao,

3

9

6 12 15 18 MONTHS FOLLOWING APPLICATION GUANICA

21

24

Figure 5. Percentage of woody defoliation in the dry, Guanica Forest site of Puerto Rico after treatment with three herbicides applied to the soil.

The rainy season usually occurs during July to October. Most of the woody plants grow vigorously during the rainy season and are deciduous during the long dry season. Refoliation on plots treated with 27 lb/A of dicamba attests to its herbicidal ineffectiveness at the Guanica site. The lack of refoliation of woody plants on plots treated with picloram and prometone is indicative of their effectiveness.

In this study, sufficient rainfall occurred after treatment to leach the herbicides into the soil and prevent large losses from volatilization and photodecomposition. The highest concentration of a herbicide in the soil profile was consistently found at the low rainfall Guanica site. On the other hand, the lowest concentrations of herbicides occurred at the continually moist Luquillo

site.

Persistence was related to the amount of herbicide applied, but the effectiveness of a herbicide on woody plants was not related to its persistence. Fenac, the most persistent herbicide, was ineffective for defoliating woody plants at all test sites. Prometone was more persistent than was picloram but effectively defoliated plants only at the Guanica site. Picloram effectively defoliated woody plants at all three test sites.

Although a high degree of woody plant defoliation was obtained from several treatments, total vegetation control was short-lived. Secondary succession occurred within 18 months on all defoliated plots at all test sites. Grasses, herbaceous plants, and vines generally were more numerous than were woody tree seedlings.

The amount of rainfall and increased light penetration appeared to influence secondary succession more than did the herbicidal treatment. The number and density of successional species were greatest on the wet Luquillo site and smallest at the dry Guanica site. In general, the number and density of successional species were greater on plots that had been defoliated. This suggested that