duction over the 12-month period-including livestock-ranged
in value from EC$66 to EC$33,500 (EC$2.70 x US$1.00). Cash
incomes (gross) from crop sales varied from nil to over
EC$32,000, with 54 of the 87 farmers earning less than
EC$2,000. Only 16 farmers earned more than EC$4,000, and on-
ly five of these more than EC$10,000. Of course, some of the
farmers with low crop incomes were livestock farmers. Twenty of
these 87 farms sold less than 50% of their total production for
cash; 42 sold more than 75% for cash. Output:input ratios (i.e.,
the ratio of the total value of production to the cash cost of inputs
used) ranged from less than one to infinity, the latter indicating
no cash expenditure on inputs. Twenty farmers had a ratio ex-
ceeding ten, and 44 had ratios less than five.

Production Systems
 A preliminary study (part of CARDI's current FSR/D Project)
 has identified over 60 crop-based production systems, some of
 which include a livestock component, among the seven countries
 of the East Caribbean. Thirteen of these are banana-based, ten
 are vegetable-based, and seven are sweet potato-based. Some of
 these may prove -on further study -to be only minor variants of
 others, but these figures serve to illustrate the diversity of the
 systems with which CARDI is working. The systems include inter-
 or mixed-crop systems (sometimes of crops of diverse form and
 longevity), pure stand systems, and rotations or sequences of
 crops. Some systems represent points in a continuum from a
 multi-crop mix to a system with one or two dominants.
 Thus, there is great diversity in all aspects of small scale farm-
 ing. Significant--in terms of developing improved technolo-
 gies-are the diversities in:
 1. emphasis on sales, cash incomes (and cash flows);
 2. the level of purchased inputs used, and in the ability and
 willingness to purchase inputs;
 3. production systems, especially at the "micro" level--the
 crop mixes and planting patterns for example;
 4. agro-ecological components, especially rainfall patterns
 and dry season duration;
 5. weed floras, consequent upon the agro-ecological com-
 ponent, the production systems, and methods of pro-
 duction.


 Weed Control Methods
 Present, or traditional, methods of in-crop weed control rely
 heavily on manual labour.

 Physical Methods
 1. Topping, by cutlassing, at 50 70 cm, to facilitate reaping
 of cocoa, coconuts, etc., may be employed. The effects on
 weeds are only temporary.
 2. Cutlassing to a stubble is common in bananas and tree
 crops and develops over time a sward of perennial weeds.
 3. Clean-weeding by cutlass, knife or sickle, cutting close to
 the soil surface, is used in short-term crops, especially
 vegetables.
 4. Hoeing for weed control and moulding-up is used in short-
 term crops such as peanuts and sweet potatoes. Some crop
 damage and root-pruning may occur.
 5. Hand-pulling is used mainly in vegetables and in
 nurseries.
 6. Mulching-usually with grass-is done in some dry areas
 both to control weeds and conserve moisture.


Biological and Cultural Methods
 1. Grazing may be permitted, mainly under established


coconuts.


134


 2. Intercropping may contribute to weed control, but often
 prevents the use of herbicides, and may make manual
 weeding more difficult. It does, of course, increase land
 utilisation.
 3. Cover crops or live mulches are rarely used, but viny
 legumes have been used in the past (Hammerton, 1984b).

 Chemical Methods
 1. Gramoxone (paraquat) is used as a directed or shielded
 spray in row crops, bananas, tree crops and in land
 preparation.
 2. Residual (soil-acting) herbicides are mainly used in
 bananas and tree crops. A few (e.g., Dacthal [DCPA],
 Enide [diphenamid] and [formerly] Tok [nitrofen]) are
 used in vegetables and food crops.
 Herbicides are invariably applied as sprays, and water may have
to be carried a considerable distance, often uphill.
 Usually a sequence of methods is used, and herbicides normal-
ly require supplementary manual weeding for resistant weeds.

Land Preparation Methods
 This activity also relies heavily on manual labour. Nineteen of
the 79 farmers characterized (Hammerton, 1984a) used tractor
services for basic land preparation, but this did not obviate the
need for further manual work. Manual land preparation typically
comprises the following sequence of operations:
 1. cutlassing of weeds which are left to dry, heaped and
 burnt, or may be buried in mounds or ridges (Paraquat
 may be used before cutlassing);
 2. forking, or tilling with a heavy hoe, to loosen the soil, bury
 residues and destroy live weeds;
 3. bed, ridge or mound formation by fork, spade and/or
 heavy hoe.
 For some crops, notably bananas and tree crops, minimum
tillage based on paraquat and/or cutlassing may be used, but
tillage will be necessary if intercrops are to be planted.
 The amount of weed growth present at the start of land
preparation depends, inter alia, on the duration of the rainfall
during the fallow period and the use made of the fallow, if any. It
may comprise woody scrub, tall grasses and vines after a fallow of
a year or more. It may comprise a dense, short, grazed sward of
mainly perennial grasses if it has been grazed. Or it may comprise
a sparse growth of grasses and broad-leaved weeds if the fallow
period was short and dry, therefore, the amount of labour re-
quired for land preparation varies widely. Few, if any, small
farmers clean cultivate between cropping seasons, even though
this would reduce subsequent weed problems. The weed fallow is
an important forage resource for ruminants, and helps to restore
soil fertility, and minimise erosion. It does ensure the prolifera-
tion of weed seed and vegetative propagules.

Constraint to Change
 Weed control is often poor on small farms: control is delayed
into or beyond the critical period of weed competition. The small
farmer, however, must optimise the use of his resources over the
entire farm -often over several crop production systems. If hired
labour is difficult to obtain when required and/or cash is not
available to pay hired labour nor to purchase herbicide, then the
farmer may be justified, for example, in:
 1. concentrating on land preparation and planting rather
 than weeding crops already planted;
 2. neglecting weed control in hardier or lower value crops in
 favour of the more weed-sensitive and higher value crops;
 3. allowing weeds to grow in later banana ratoons, or among
 tree crops, to ensure forage for livestock;
 4. intercropping to maximise production from a limited land
 area, even though this makes manual weed control more


time-consuming and may preclude herbicide use.

 PROCEEDINGS of the CARIBBEAN FOOD CROPS SOCIETY-VOL. XX