islands confirms the production patterns observed above, i.e.,
that small farms account for a majority of the food crop,
vegetable and fruit and nut production (Bourne and Weir,
1980).
 Determining the relative efficiency of small farms was plagued
not only by census data limitations but also by the unavailability
of acceptable local estimates for allocating acreage to specific
crops and livestock rearing in the highly mixed intercropping
systems characteristic of the USVI. These factors precluded the
utilization of more sophisticated efficiency criteria commonly
employed elsewhere to assess productivity in mixed farming (Har-
wood, 1979). As a result, two crude measures were constructed:
output and/or livestock per acre and output per tree of bearing
age.
 Tables 8 and 9 present the results. In the first case, relative effi-
ciency was estimated for 1983 by dividing the total output pro-
duced by each farm size for each fruit/nut selection by the respec-
tive acreages in fruit/nut production for each farm size classifica-
tion. According to this method, the smallest scale of under 3 acres
was most efficient, achieving the highest production per acre in
every fruit/nut category. A similar analysis of livestock productiv-
ity-total number of sheep/goats/hogs/cattle divided by total
acreage in pasture and grazing land-generated similar results.
With the exception of bananas, coconuts, and grapefruits, 3-9
acre holdings were second in efficiency. However, although these
findings do capture the intensity of effort on the two smallest-
scale classifications, they should be accepted guardedly because of
the aggregative nature of the methodology, which ignores inter-
cropping patterns and variations in land quality, and because of
the assumption of constant output quality, especially with respect
to livestock, across farm size categories.
 In the second experiment, the ratios of harvested fruit/nut
production to respective trees/hills of bearing age were calculated
for each farm size for only two years for which census data were
available, 1975 and 1983. Although the figures in Table 9 in-
dicate some large productivity differences for the same farm sizes
across the two years-perhaps due to topographical variations,
tree stock maturity differentials, and/or the vagaries of
weather-the overall results generally suggest that sizes of under
3 acres and 20-49 acres were relatively most efficient in non-citrus
and citrus products respectively, while units of 10-19 acres were,
with some exceptions, least efficient.


CONCLUSIONS
 In summary, these analyses of relative productivity, confined
 by the limitations of crop selection imposed by census data, sug-
 gest that in terms of gross output contributions to the territorial
 economy farms of 3-9 acres were generally superior in small
 livestock and non-citrus fruit/nut production while farms of
 20-49 acres were superior basically in citrus produce and sheep
 and cattle. In terms of relative efficiency or production per acre
 and per tree/hill of bearing age, farms of under 3 acres were most
 efficient across all tests with differentials clearest in non-citrus
 fruit/nut products while farms of 20-49 acres demonstrated their
 comparative advantage in citrus. In all cases, farms of 10-19 acres
 scored the lowest performance.
 Such findings should assist policy-makers in view of:
 1. The resource constraints that circumscribe such efforts and
 call for prioritizing;
 2. Escalating USVI food imports which have risen from $5
 million in 1960 to over $80 million presently (Government
 of the Virgin Islands, 1980:26);
 3. The long period of USVI agricultural decline; and
 4. The common problems infesting agriculture here and
 elsewhere in the region (Belisle 1983).


However, because of the rudimentary nature of the analysis and
numerous data gaps, the primary implication of the study is to


VOL. XX-PROCEEDINGS of the CARIBBEAN FOOD CROPS SOCIETY


TABLE 7. Distribution of selected fruits/nuts production by small-farm size. U.S.
Virgin Islands, 1982.

 Under 3 ac. 3-9 ac. 10-19 ac. 20-49 ac. Total

 % Avocados 21.7 35.3 9.3 21.9 88.2
 % Bananas (bunches) 17.1 29.5 17.0 6.1 69.7
 % Coconuts 10.7 24.0 12.4 22.9 70.0
 % Grapefruits (Ibs.) 17.3 16.0 10.0 32.6 75.9
 % Limes/Lemons (Ibs.) 23.3 25.6 7.9 38.1 94.9
 % Mangoes 8.9 25.8 5.0 17.1 56.8
 % Oranges (Ibs.) 11.6 37.9 3.5 28.4 81.4
 % Papayas (Ibs.) 33.1 30.1 11.8 18.6 93.6

 SOURCE: U.S. Census of Agriculture for the Virgin Islands, 1982. Bureau
 of the Census, Washington.


TABLE 8. Selected fruits/nuts and livestock per acre by small farm size. U.S.
Virgin Islands, 1983.

 Under 3 ac. 3-9 ac. 10-19 ac. 20-49 ac.
 Total acres in
 fruits/nuts 38 170 65 148
 Output per acre:
 Avocados 182 66 46 47
 Bananas (bunches) 52 20 30 5
 Coconuts 51 26 34 28
 Grapefruits (Ibs.) 21 4 7 10
 Limes/lemons (Ibs.) 77 19 15 32
 Mangoes 492 319 160 243
 Oranges (Ibs.) 19 14 13 12
 Papayas 42 9 9 6
 Total acres in pasture
 and grazing land 22 217 182 506
 Total cattle, sheep,
 goats, and hogs 1,564 2,735 952 1,995
 Total livestock per acre 71 13 5 4

 SOURCE: See Table 7.




TABLE 9. Ratios of harvested output to trees of bearing age for selected
fruits/nuts. U.S. Virgin Islands, 1975 and 1983.

 Under 3 acres 3-9 acres 10-19 acres 20-49 acres

 Avocados
 1983 29.0 22.3 17.4 44.8
 1975 38.7 14.8 25.9 12.1
 Bananas (bunches)
 1983 0.55 0.44 0.16 0.38
 1975 0.70 0.54 0.43 1.29
 Coconuts
 1983 9.4 10.3 3.7 5.4
 1975 22.8 10.4 7.5 7.8
 Grapefruits (Ibs.)
 1983 22.2 10.3 4.9 31.3
 1975 7.1e 2.3 7.9 10.7
 Limes/Lemons (lbs.)
 1983 12.0 7.3 8.9 22.4
 1975 24.6 15.4 13.5 25.0
 Mangoes
 1983 82.0 47.7 34.6 94.4
 1975 61.4 66.0 19.4 8.7
 Oranges (lbs.)
 1983 12.3 15.0 8.8 15.0
 1975 7.1 7.8 2.6 10.5
 Papayas
 1983 8.6 2.7 2.2 7.8
 1975 8.5 4.5 5.4 3.4

 SOURCES: U.S. Census of Agriculture for the Virgin Islands, 1974 and 1982.
 Bureau of the Census, Washington.
 eEstimated