TABLE 2. Foliar-applied herbicides', classified according to their requirement for
fine dry weather following application.

 Less than two hours required Four hours or more required
 2
 ("Low risk") ("High risk")

 Fusilade (fluazifop-butyl) Asulox (asulam)
 Gramoxone (paraquat) Basagran (bentazon)
 Reglone (diquat) 2,4-D
 Round-up (glyphospate)
 Talent (asulam + paraquat)



 'Inclusion of a herbicide in the above Table and mention in the text, does not con-
 stitute an endorsement. Nor does omission imply criticism.
 2Within the "Low risk" group Gramoxone and Reglone carry less risk than Fusilade,
 as they require only about 30 minutes of fine weather, compared to about 1.5 hours
 for Fusilade.


 Furthermore, good weed control may only be justified if inputs
of fertilisers, nematicides, etc., are near-optimal.
 Herbicides can benefit the small farmer by easing the labour of
land preparation, and extending the critical period of weed com-
petition, even if the control obtained is only partial. The farmer
may therefore be able to manage a greater area of cultivated land
with herbicides (Hammerton, 1974). But herbicides have several
constraints. Many are susceptible to rainfall either, in the case of
foliar-applied herbicides, by washing-off, or in the case of soil-
acting herbicides, by leaching. Tables 2 and 3 classify some com-
mon herbicides in terms of susceptibility to rain. Clearly the risk
of washing-off or leaching varies with the rainfall pattern. "High-
risk" herbicides stand a higher chance of failure during seasons of
heavy, frequent rains, than "low-risk" herbicides. Weed growth is
more rapid during seasons of high rainfall. Cost is another con-
straint. A farmer may require only a few fluid ounces, but may
need to buy a litre at a cost of EC$100-200. Cost per unit area can
be reduced by spot treatment, but with labour rates ofEC$15-20
per day, manual weeding may be a better proposition-provided
labour is available. No weeds are resistant to manual methods.
Partial budgetting is necessary before recommending the in-
troduction of a herbicide-but a holistic or whole-farm perspec-
tive is essential (Hammerton, 1984b). Paraquat (formulated as
Gramoxone in the East Caribbean) currently sells for about
EC$16-17 per litre, and although ineffective against many weeds
-vines and perennials, for example-it is easily the most cost-
effective and attractive herbicide for most small farmers. Advan-
tages are its "rain-fastness" and its rapid and visible action.


A Systems Approach
 CARDI, under its current FSR/D Project, is engaged in
developing and testing improved weed control systems for small
farmers. Attention is being focused on those production systems
rated high priority in terms of contribution to domestic food sup-
ply and potential for export.
 Improved systems should satisfy some, at least, of the follow-
ing criteria:
 1. integrated systems, utilizing several components;
 2. integrate into existing production systems with minimal
 disruption, or be an integral part of an improved total
 technology;
 3. require levels of expenditure realisable by the target
 farmers;
 4. economically sound, as shown by partial budgetting,
 cost:benefit ratios, etc.;


TABLE 3. Soil-applied herbicides', classified according to ease of leaching in soil.

 Not readily leached Readily leached
 ("Low risk") ("High risk")

 Dacthal (DCPA) Alanap (naptalam)
 Enide (diphenamid) Amiben (chloramben)
 Gesagard (prometryn) Dalapon
 Gesaprim (atrazine)
 Lasso (alachlor)
 Lorox (linuron)
 Maloran (chlorbromuron)
 Probe (methazole)
 Randox (CDAA)

 'See note to Table 2.


 5. ecologically sound, especially the herbicide component, if
 any;
 6. minimal risk of failure associated with soil and weather fac-
 tors.
 The components include:
 1. cultivations for land preparation;
 2. herbicides for land preparation;
 3. post-planting cultivations (e.g., hoeing, cutlassing, etc.);
 4. post-planting herbicides, directed, overall or spot;
 5. live or dead mulches
 6. intercropping
 7. "managed" weed swards.
 The programme of work comprises both component testing
 and evaluation of systems. Component research includes her-
 bicide evaluation, testing of herbicide wipers and of wheel hoes
 and hand tools, etc. System evaluation is the testing of systems
 which may include land preparation herbicides and methods,
 pre-emergence herbicides, and sequences of post-planting her-
 bicides and/or physical methods of weed control. Cost is a major
 consideration in selecting systems for evaluation. The programme
 includes field station experiments and on-farm testing and
 evaluation. Basic to this programme is a weed control survey at
 present underway in several countries.
 As a logical basis for developing weed control systems, and
ultimately recommendations, four "domains" are recognized
(Table 4). Generalised systems suggested for these domains are as
follows:
 High rainfall- wet season Manual methods give only short-
term control. Intercropping may suppress weeds, if weeds are
known to be sparse, but may complicate control under very
weedy conditions. Low risk foliar applied herbicides, especially
Gramoxone can be used, but control will be temporary. Fusilade
can be used as a spot treatment and over the top of broad-leaved
crops for grass weeds. Manual methods must be used for resistant
weeds, especially woody species. Low risk soil-acting herbicides
should be avoided, unless there is adequate ground cover from
the crops and/or intercrop(s). High risk soil-acting herbicides
should not be used.
 High rainfall-dry season Manual methods are moderately
effective. Round-up can be used judiciously for land preparation,
and spot treatment of perennial weeds. Gramoxone can be used
as a management tool to maintain a cover of perennial grasses on
steep slopes. Fusilade can be used for ring-weeding control of
perennial grasses. Gramoxone should be used for annual weeds.
Low risk soil-acting herbicides can be used at the start of the dry
season on flatter lands.


VOL. XX-PROCEEDINGS of the CARIBBEAN FOOD CROPS SOCIETY


135