Geoforum, Vol. 22, No. 3, pp. 333446, Printed in Great Britain
0016-7185/91$3.00+0.00 @ 1991 PergPnm Press pk
1991
Identification of Marginal Agricultural Areas in Ontario, Canada
BARRY SMIT,* Guelph, Canada, JUSTINE BRAY,? Auckland, New Zealand, and PHILIP KEDDIE,* Guelph, Canada Abstract:
This paper develops an approach to identifying marginal areas for agriculture based on the principle of land-use viability. A working model is derived from the classical laud-use theories of von Thunen and Ricardo. The concept of landuse viability relates to the ability of land to generate positive economic returns consistently. Economic returns are specified for agricultural uses on various land types under a range of economic and environmental conditions. The distribution of returns serves as a base to define marginal land and, subsequently, the marginality of broad areas or regions. The practicability of the approach is tested via an empirical investigation for the province of Ontario, Canada.
Introduction
Areas in agriculture which have limited productive potential due to unfavourable biophysical and/or socio-economic conditions are frequently described as marginal, a term which inherently implies disadvantage relative to other areas. These regions of agricultural disadvantage are of interest in geographic study from at least two perspectives. Regions at the margin of agricultural production are often associated with rural underdevelopment because of modest returns to agriculture. These areas may experience a cumulative process of decline resulting in farm abandonment and the gradual deterioration of the farm servicing network due to lack of patronage. As the agricultural infrastructure is depleted farming becomes increasingly difficult for the remaining operators, reinforcing the process (BEATTIE et al., 1981). Consequently these areas are frequently the target of initiatives to boost regional development in an attempt to ameliorate problems of rural disadvantage. A second perspective
pertains
to activities at the
*Department of Geography, University of Guelph, Guclph, Ontario, Canada NlG 2Wl. tKrta Associates, Newmarket, Auckland, New Zealand.
agricultural margin, which are often only minimally viable and sensitive to even modest changes in biophysical, social, and economic conditions. Thus, stresses on agricultural systems are often first revealed by changes at the margins because these areas act as barometers for the general health of the agricultural sector. Any practical investigation of these hypotheses about the sensitivity or underdevelopment of marginal areas requires an operational definition of marginality. The relatively sparse literature on the subject presents a confusion of terms and little systematic conceptualization and definition. This paper aims to clarify some of the terminology, to distinguish the various concepts of marginality, and to develop an approach to identify land which is marginal for agriculture. The approach is then applied to the province of Ontario, Canada.
Biophysical
and Socio-economic
Marginality
Marginality has been used to describe a wide range of situations in agriculture with many terms used interchangeably (BEATTIE et al., 1981). For example, an area described as marginal due to biophysical limitations of the land base is different from a farm unit 333
334 which is economically marginal. This section reviews the ways in which biophysical and socio-economic conditions have been interpreted as contributors to marginality in agriculture. Studies examining the production of particular agricultural commodities often acknowledge the limitations imposed by biophysical qualities of the land (FRANCIS, 1970). Consideration of biophysical limitations on agriculture is a long-established theme agricultural geography (BAKER, 1921; CAGES, 1949; MCCARTY, 1954; REEDS, 1964). Analyses of agricultural location have traditionally isolated factors of the biophysical environment to explain spatial patterns (PACIONE, 1986). Similarly, studies exploring the limits for agricultural activities have concentrated on identifying the nature and degree of biophysical constraints. ZIMMERMAN (1951) proposed temperature, moisture, topography/landforms, and soils as the four ‘frontiers’ of agriculture. TROUGHTON (1977) cites terrain and soil, while IRONSIDE et al. (1974) argue that climate is the most important limiting factor in their respective studies. However, the determinants of the biophysical margins of agriculture are essentially crop-specific, as ILBERY (1985) demonstrated in his review of the effects of climatic conditions on cereal crop yields. Obviously, any consideration of bioptysical limits on specific crops or agriculture in general must recognize attributes of land and climate together with their interactions (GRIGG, 1984). Despite increasing technological abilities to modify and adapt to the biophysical conditions, their influence on the spatial structure of agriculture remains important at the regional level (ILBERY, 1985). A biophysically marginal environment is generally interpreted as one in which agricultural crops do not grow well consistently, and/or which requires major inputs to achieve reasonable levels of production. Areas of agricultural production (and, similarly, marginal areas) can only partially be deciphered in biophysically deterministic terms as agriculture is also a social and economic activity (GRIGG, 1984; PACIONE, 1986). Economic and social conditions can enhance or reduce opportunities for agriculture regardless of the state of the biophysical environment [for example MAXWELL (1966) and VARJO (1979)]. So marginality may reflect socio-economic constraints on agricultural viability. Low returns to investment may be associated with paucity or disadvantage in one or more of the factors of production. Socio-economic factors affect the viability of a farm unit, which is dependent upon whether rates of return
Geoforum/Volume
22 Number 3/1991
fall above or below some accepted minimum (SCFI, 1969). An economic interpretation focuses on farm ownership and operation, and on how characteristics of the land together with other inputs influence the nature of individual farm units and the condition of agricultural regions. It is generally accepted that the level of farm income is related to the level of farm management (OECD, 1984; SCFI, 1969; ZIMMER and RODD, 1971). Farmers’ education levels and skills play a role in overall farm viability (MANDALE, 1984). Lack of capital can also significantly constrain agricultural viability. The opportunities to accumulate capital resources are restricted for farm operators in some areas because of low profit levels, poor credit facilities, or poor borrowing ability, often related to low farm values (COMEAU, 1974). Marginal regions are sometimes characterized on the basis of low levels of intensificatiomand capitalization of farming operations (BEATTIE et al., 1981). TROUGHTON (1977) suggests that marginal areas reflect disadvantage in both economic and social conditions which may arise from isolation, the macroeconomic environment or physical conditions which impact negatively on agricultural opportunities. These regions can be identified by such characteristics as an aging population and low capital commitment with consequent regional responses in the form of offfarm work, rural depopulation, farm abandonment, and persistent rural poverty. During the 1950s economic decline and land abandonment in marginal areas of Canada resulted in increased social problems such as high infant mortality, low income, and high unemployment (TROUGHTON, 1977). Such conditions have been described by FRANCIS (1970) as social marginality.
Marginal
Lands, Farms, and Areas
The term marginality has been applied to a wide range of spatial situations in the agricultural context, and is subject to various interpretations. McCUAIG and MANNING (1982) define the location of the margin relative to distance from an urban centre. MANNING and BARDECKI (1988) use this definition to identify distinct belts of marginal agricultural land in western and eastern Canada, typically located at the agriculture-forest interface. CHAMPION (1983) locates marginal agriculture between the wildscape and the farmscape, a region characterized by a mixture of productive rural land and wilder-
Geoforurn/Vcrlume
335
22 Number 3/1991
ness areas, where human activity has little or no impact. TROUGHTON (1977, p. 7) also defines marginal land from a locational viewpoint as a zone “ . . . almost everywhere peripheral to or beyond the core region of settlement . . .“. This area exhibits disadvantage relative to more settled areas not only with respect to distance from the core, but also in economic and social conditions. SYMONS (1970, p. 83) argues that the term marginal is often used to describe lands of low productivity, while genuine marginal land only occurs when the use that occupies it fluctuates, or “. . . is likely to be abandoned from all uses because it is poor”. He suggests that definition becomes easier as the emphasis moves to the financial aspects of the farm unit and profitability. BEATIIE et al. (1981, p. 3) present a definition of the agricultural margin as “. . . a line . . . which separates the area wherein agricultural production is economic, from that part where it is not”. Accordingly, marginal lands are designated as “ . . . those lands which at any point in time are at or near the margin for agriculture” (BEA’ITIE et al., 1981, p. 3). Economists speak of some land areas as being marginal for particular types of use, the usual inference being that the areas are at or close to the norent (zero net returns) level for the particular uses considered (BARLOWE, 1978). This brief review suggests a basis for clarifying terminology as it is applied to marginal lands, marginal farms, and marginal areas or regions. The different concepts are illustrated in Figure 1. Marginality relates fundamentally to the economic viability of land uses. The viability of any use on any land parcel is determined by many factors including biophysical, locational, economic, and social conditions. Marginal lands (Figure 1) are land parcels or land types for which there are few viable uses, or for which uses are
I
VIABLE IAND USES
either only just viable or not always viable. These are distinct from marginal farms (Figure 1) which refer to actual operating units which are only just viable or not always viable. SYMONS (1970) describes a marginal farm as one unable to yield a regular or satisfactory profit. A farm may be marginal because of a poor land base, but marginal farms can also exist on superior lands. Marginal farms may exist because of a variety of factors unrelated to inherent land quality, such as farm size, the managerial ability of farm operators, or changes in price or policy conditions. By the same token, off-farm conditions, including employment, may result in operations which are sustained despite an inferior land base. Marginal areas or regions can be defined either on the basis of lands or of farms. Marginal (land) areas (Figure 1) are regions which are composed of significant proportions of marginal lands. Marginal cfarm) areas are defined according to the regional distribution of marginal farms. The remainder of this paper focuses on marginal lands and marginal land areas. The intent is to define lands according to their potential to be viable for agricultural uses. An explicit definition of marginality is established as a basis for identifying marginal lands and subsequently marginal areas.
Marginality and Land-use Viability
Land-use viability implicitly underlies most discussions of marginality. For an agricultural use to be viable on a given parcel of land, it must be able to generate positive returns consistently. Economic returns to land are at the heart of classical theories of land use. RICARDO (1817) focused on soil fertility as a factor affecting economic returns and hence landuse competition. Von Thunen (HALL, 1966) highlighted the way transportation costs influenced variations in returns and hence patterns of land use. These theories propose that a range of biophysical and socio-economic factors in combination influence variations in returns (rent) from the land. The general form of these economic models of land use can be presented as ER, = Y,Pi - C,
Figure 1. Concepts of marginality in agriculture.
where ER, = economic returns from use i per area of land parcel i, Yii = output from use i per area of land parcel i, Pi = market price per output of use i, and Cii = cost of production marketing (including transportation costs) of use unit area of land parcel i.
(1)
unit unit unit and i per
336
GeoforumNolume
Equation (1) includes the major determinants of land-use returns and viability: yield, market price and costs of production. Both biophysical and socioeconomic constraints on agricultural production are captured. The joint impacts of biophysical elements (e.g. climate, soil, topography) are reflected in the yield values (Yii>, and in the costs of production (C,), both of which vary depending on environmental conditions. Production costs are also influenced by socioeconomic conditions, including costs of fixed and variable inputs, technology, and skills. The demand or market for commodities from a use are reflected in the product price (Pi). Likewise, external influences such as government support, commodity pricing policies, subsidies, and even trade controls should be captured in production costs (Cij) or product prices Cpi)
*
For a given use (i) on a given parcel (j), viability is a function of the ERs: Vij = f(ERij)
= f( Yq, Pi, C,>
(2)
where Vii = viability of use i on land parcel j. This model can be used to establish the viability of any land parcel for any use. Obviously the empirical implementation is not a simple matter, as questions pertaining to year-to-year variation in yields, prices, and costs, and issues relating to the on-farm use of products need to be resolved. These questions will be addressed subsequently; the main point here is that it is possible to define a viable land use according to consistently positive ERs, which are in turn a function of the biophysical and socio-economic environments.
Defining
Marginal
22 Number 30991
Ma = fa(Mj) for all the land parcels j within an
area a
(4)
where M, = marginal land area. Specific decision rules for identifying marginal lands and areas (i.e. the form of functions h and fo) are addressed in the empirical application which follows. While this and subsequent sections focus upon marginal land and (land) areas, similar procedures could be employed to identify marginal farms and (farm) areas.
Empirical Application
to Ontario
The remainder of this paper reports upon an empirical application of the approach to identify marginal areas in the province of Ontario. A test in this study area was feasible because pertinent data already exist on land uses, land types, and agricultural yields, prices and costs, and the level of detail is sufficient for demonstration purposes.
The analysis begins by assessing the viability of agricultural uses across land types. The viability of a given use is a function of the ERs which will accrue to a land parcel from that use. Since yields and production costs are influenced by larger-scale economic and climatic characteristics, as well as by land type, a regional component (r) is added, and the operational equation for returns becomes E&r = Yi,,pi - Citr
(5)
where ERi, = economic return from use i on land type tin region r, Yih.= yield of use i on land type tin region r, Pi = price for use i, and Ci, = cost of production of use i on land type t in region r.
Lands and Areas
Marginal lands are here defined as land parcels or types for which there are few viable uses, or for which uses are either only just viable or not always viable. Thus, they are defined according to the lack of certainty of positive returns, which in turn is established by considering the viability of all land uses on the particular land parcel: Mj = fj( Vij) across all uses i
(3)
where Mj = marginality of land parcel j. Marginal (land) areas are defined here as regions which have a high proportion of marginal lands @thin their boundaries. To identify marginal areas, the marginal lands defined using the viability indices [equation (3)] must be aggregated:
Data Regions (r). Eleven regions (r) are identified within
Ontario based on climatic, economic, and administrative features (Figure 2). These were derived by first ‘delineating the province into climatic zones based on crop-sensitive thermal conditions as measured by corn heat units (BOND et al., 1981). Seven climatic zones with similar agroclimatic properties (e.g. growing season length, precipitation, temperature) were identified. To facilitate comparisons with existing data sources the climatic boundaries were matched to township and census division boundaries. Ontario was also divided into six major economic/administrative areas to capture regional differences in production characteristics and costs.
Geoforum/Volume
22 Number 30991
These divisions were superimposed over the climatic zones with an end result of 11 distinct regions (LEG, 1983). Land types (t). For the land cleared and available for agriculture in each region identified above, up to seven land types are defined on the basis of their biophysical characteristics which relate to crop productivity (Table 1). The soils in land types A, D, and G are generally regarded as having a high productive capacity, whereas those in categories B and C have a moderate to low rating. Soil conditions on land types E and F significantly restrict the production of horticultural and field crops. Uses (i). Fourteen agricultural uses representing the major crops in Ontario are included in the analysis: grain corn, silage corn, oats, barley, winter wheat, hay, soybeans, white beans, apples, grapes, peaches, sweet corn, peas, and field tomatoes. Together these uses accounted for 51% of the available farmland in 1981, and occupied 85% of the cropland area. Mixed
337 grain, occupying 8.7% of the cropland area, is not included, but its position should be effectively captured by oats and barley. Yields (Y) . Estimates of current productivity for each use on each land type within each region are available from published sources (LEG, 1983). These values, measured in tonnes/ha, are derived from a crop productivity model which utilizes data on growing season length, thermal energy, seasonal rainfall, and soil moisture-holding characteristics (AES, 1982; LEG, 1983). Prices (P) and production costs (C). Data on the prices and production costs for 13 of the agricultural uses were obtained from HASLETT (1983). Commodity prices ($/tonne) are based on data for 1977-1982. Costs ($/ha), corrected to 1982, included hired labour, machinery and equipment, materials, marketing fees (where applicable), crop and liability insurance, and interest on operating costs. Costs of
Geoforum/Vohune
338
22 Number 3/1991
Table 1. Land types*
Land type
Biophysical characteristics
A
Well-drained loamy soils Relatively level topography Fine-textured clays Moderately well to well-drained Coarser-textured sands and gravels, or shallow soils (less than 1 m to bedrock) Well to excessively well-drained Imperfectly drained soils which have been tile-drained Imperfectly drained soils which have not been tile-drained, and all poorly to very poorly drained soils Extremely stony or shallow soils Well-drained stony or shallow soils Rolling to hilly topography
B C D E F G
*Source: LEG (1983).
hay production, not available in Haslett, were estimated using the same techniques. The effects of differential land cost on production costs (and hence on returns) are not captured in this formulation. It is possible that the methods used to calculate costs of production have underestimated the variation in ERs among regions. However, these available cost estimates are based on a consistent procedure suitable for the level of aggregation employed in this empirical analysis. Data on yields, prices and costs are compiled for each combination of use, land type, and region. With 14 uses, seven land types, and 11 regions, economic returns could in theory be calculated for up to 1078 combinations, but, in practice, not all land types occur in all regions.
Variability
in Climate and Price
Land-use viability is defined as the ability of a use to generate positive economic returns consistently on a given land type. The data described above depict returns given long-term weather conditions and prices averaged over several years. As a consequence, the returns are an indication of the viability of a land use under only those average conditions. In order to account for the range of economic and biophysical conditions likely to be encountered, variability in both yield and price should be considered. This permits definition of viable land uses on the basis of probable returns, reflecting the likelihood of returns over the range of conditions farmers must face from year to year. Crop yields can be affected
significantly by avail-
ability of moisture. In this study three possible precipitation levels are considered. A year with average precipitation for the province would have 7.6 cm of rainfall per growing season month (i.e. MaySeptember). In a wet year this would increase to 10.2 cm, and decrease to 5.1 cm in a dry year (BOND et al., 1981). To estimate the probability of each of these three conditions occurring, historical precipitation patterns were analyzed (DEPARTMENT OF TRANSPORT, 1959-1976; ENVIRONMENT CANADA, 1977-1982). For three climate stations in each region, daily rainfall totals were compiled for a 25-year period from which the monthly means were calculated for each region. From these frequencies the probability for each region of average, wet, and dry years occurring were determined (Table 2). HASLETI (1983) provides data on average crop prices as well as prices one standard deviation above and below these means. Comparable data were compiled for hay, which was not included in Haslett’s analysis. The probability of being 1 SD away in either direction in a normal distribution is 0.1587. The probability of the mean price occurring, or more Table 2. Precipitation
probabilities
Region
Wet
Average
Dry
NO1 E03 co2 co3 wo2 wo3 SC04 SW04 SW05 SW06 SW07
0.28 0.22 0.21 0.22 0.20 0.22 0.11 0.18 0.20 0.20 0.18
0.64 0.62 0.70 0.64 0.66 0.54 0.72 0.68 0.62 0.63 0.65
0.08 0.16 0.09 0.14 0.14 0.24 0.17 0.14 0.18 0.17 0.17
Geoforurn/Volume
22 Number 3/1991
339
precisely the range between the + 1 and -1 SD in price, is 0.6826. For each combination of land use, land type, and region we now have nine scenarios and their associated probabilities (three for precipitation and three for price). As it is clearly not possible to present the full data set in this paper, Table 3 provides a sample of information relating to ERs for a selection of uses (4) on one land type (A) in one region (SW06). Yields per hectare are provided for each crop given dry, average, and wet years. The average price per tonne (P2), as well as the price 1 SD deviation below (Pl) and above (P3) this mean, are also reported, as are production costs. From this information it is possible to calculate the ER expected under each of the scenarios. To determine the likelihood of specific combinations of price and precipitation occurring, joint probabilities (the products of price and precipitation probabilities) are estimated. These probabilities sum to 1. An adjusted economic return (AER), which reflects both the estimated return under a given scenario and the probability of that scenario occurring, can be calculated:
AERS;f=ER$$PR$-$
Probable Economic Returns and Viable Uses
Once the AERs are determined, the long-term probable economic return (PER)for a use on a land type in a region is calculated by summing over all the scenarios:
cc PC
AERG.
Following the procedure outlined above it is possible to identify all viable use/land type/region combinations for Ontario. Figure 3 indicates significant contrasts between the regions of the province. Hay and sweet corn are the only viable uses in region NOl, whereas all 14 uses are viable in region SW07 on at least one land type, with many uses being viable on several land types. In most regions, land types A, D, and G support the greatest number and range of uses. Hay, followed by grain corn and silage corn, appear to be the most versatile uses, as opposed to peaches which are viable only in region SW07. In all regions no crops are viable on land type F.
(6)
where p = price scenario, c = climate scenario, and PR = probability.
PER,,=
to set a subsistence level (e.g. to account for operator wages and land costs), and to define viable uses as only those above this value. Unfortunately, this option did not prove practicable as no basis could be established for suitable levels of subsistence return in Ontario as a whole, let alone for particular regions. Thus the zero PER is taken as the threshold for viability, as an indicator of the potential of the land for a specified use. As defined, a positive PER recognizes the likelihood of success over the long term given variations in precipitation and prices.
Marginal
Lands and Marginal
Areas
The marginality of lands relates to the distribution of economic returns across uses. Marginal lands might be thought of having some prospect for success (e.g. have at least one use which is viable) but where there are limited options among viable uses. Land units (land types in regions) need to be classified according to their ability to generate positive returns over the range of uses.
(7) The approach adopted is based upon the number of viable uses (i.e. positive PERs) for land units. How-
To illustrate, the PER for grain corn on land type A in region SW06 is $188.77 (Table 3). This figure indicates the viability of grain corn, which can generate a positive if variable ER under eight of the nine price and moisture scenarios. In dry years which also have a low price (Pl) there is a loss on grain corn on land type A in region SWO6, but since the joint probability of such a year is low (0.027), it does not greatly influence the PER. In effect the AERP" is each scenario’s share (proportion) of the PER.
PERs provide the basis for assessing viability. For our purposes a viable use is defined as any use for which the PER is above zero. An option exists at this point
ever, other approaches are possible and are described briefly. It would be desirable to define marginality on the basis of both the number of viable uses and their levels of economic return. This could be accomplished by treating the PER as a ratio variable, and calculating an average PER across uses for each land unit, for those uses for which PER is positive: “I?
MPP,=
PPERiJn, c i=l
(8)
where MPP, = mean positive PER for land type tin region r, PPERi,= positive PERs for use i on land
2.59 2.59 2.59
1.91 1.91 1.91
11.00 11.00 11.00
Oats Oats Oats
Soy beans Soy beans Soy beans
Apples Apples Apples
A A A
A A A
A A A
14.34 14.34 14.34
2.46 2.46 2.46
2.78 2.78 2.78
6.45 6.45 6.45
Average
15.51 15.51 15.51
P1186.0 P2 236.0 P3 286.0
P1231.0 P2 268.0 P3 305.0
2.76 2.76 2.76
7.28 7.28 7.28
Pl 89.8 P2 110.9 P3 132.3
Pl 86.0 P2 113.0 P3 140.0
Wet
2.59 2.59 2.59
Price (P) ($/tonne)
-1301.00 -751.00 -201 .OO
3347 3347 3347
-679.76 37.24 754.24
310.26 401.28 492.30
20.78 79.25 138.74
3.71 58.18 113.61
183.21 253.88 324.55
26.70 200.85 375.00
Average
-111.76 18.92 149.60
Dry
258 258 258
229 229 229
528 528 528
cost ($/ha) Dry
379.56 481.68 583.80
3.71 58.18 113.61
0.027 0.116 0.027
0.027 0.116 0.027
0.027 0.116 0.027
0.099 0.431 0.099
0.099 0.431 0.099
0.099 0.431 0.099
0.099 0.431 0.099
Average
0.032 0.137 0.032
0.032 0.137 0.032
0.032 0.137 0.032
0.032 0.137 0.032
Wet
Joint probability
98.08 0.027 294.64 0.116 491.20 0.027
Wet
-462.14 313.38 1088.36
Economic return ($ha)
*Note: for price (P): Pl = 1 SD below average price, P2 = average price, P3 = 1 SD above average price. tPER = probable economic return.
4.84 4.84 4.84
Grain corn Grain corn Grain corn
A A A
Dry
Use
type
Land
Yield (Y) (tonnes/ha)
Table 3. Sample of the data base for estimating economic returns: region SW06, land type A, four uses*
-35.13 -87.12 -5.4
4.95 29.45 8.76
0.10 6.75 3.07
-3.02 2.19 4.04
Dry
PER
-67.30 16.05 74.67
PER
30.72 172.95 48.74
PER
2.06 34.16 13.74
PERT
2.64 86.57 37.13
Average
($/ha)
-14.79 42.93 34.34 -41.26
12.15 65.99 18.68 392.38
0.12 7.97 3.64 71.59
3.14 40.37 15.72 188.77
Wet
Adjusted economic return
Geoforum/Volume
22 Number 3/1991
341
NO1 (USE) Grain Corn Silage Corn Oats Barley Winter Wheat White Beans Soybeans
ABCDEFG . 0. .
.
l
*.
Hay
.
l
.
0 0
e
.
.e
Silage Corn Oats
l
White Beans
l
Soybeans
l
Sweet Corn Peas Tomatoes
a. .
l
.
.
l
.I...
l
l
l
.
ABCDEFG . l . . .
.
e.
l
.
.
. . . . .
. .
0
wo3
ABCDEFG . . . l
. .
e
l
woz
ABCDEFC . 00 l .
Barley Winter Wheat
l
l
eoee
co3 Grain Corn
l
TYPE)
l
l
.
.
l
l
0
(LAND
.
ABCDEFC . .
.
.
Sweet Corn Peas Tomatoes Apples
co2
E03
ABCDEFG
(REGION)
. . . .
.
SC04 A B C D E F <; l . . . l 0. . . . . . . .
.
.
.
.
.
.
.
.
.
l
.
.
.
. . .
. .
0 .
.
. .
l
. . .*.*
.
. .
. .
. .
Apples Hay
bee..
.
.
l
Grain Corn Silage Corn Oats Barley Winter Wheat White Beans Soybeans Sweet Corn Peas Tomatoes
l
ABCDEFC l . . . l . . .
.
l
.
l
0. *me
l
e
0. l
l
l
l
0. .e
. .
ee.
.
l
l
.
l
l
e..e
. .
.
. .
. .
.
.
. .
l
l
.
..a. . .
.
*em..
ABCDEFG ..a. .a.. l
. l . . .
SW07
a . .
. l . .
.
l
.
. . .
. . . l
. .
. . .
. l . .e.. l e
.
0. l
. .
ABCDEFG ..*. a...
. . .
.
Apples
. . .
. .
. .
. l
. .
. ..
l
e .eme
.
l
l
.
l
l
l
l l
Peaches
.
Grapes Hay
.
SW06
.
. em.0
.
SW05
SW04 ABCDEFC . ..a
.
l
..ee
0
l
. eme.
l
.
l
l
. eoee
l
l
. .eee
. l
Figure 3. Viable uses by region and land type.
type t in region r, and n, = number of uses (i) which have positive PERs on land type t in region r. To recognize variations in the relative importance of different crops, the PER values for each use could be weighted according to the total area occupied by the crop in the province. This approach assumes a high degree of accuracy in the absolute PER values. Less assumptions about data accuracy would be required by treating the PER as an ordinal scale, and ranking the land units from lowest to highest value, again ignoring non-positive PERs. Summing the ranks across uses for each unit, and then dividing by the number of viable uses would yield an index of the
relative viability of each land unit. Land units with higher values would be expected to have a greater likelihood of success. The approach adopted here assumes the least about the data, distinguishing marginal lands simply according to the number of viable uses. With this technique the absolute value of the PER is not critical, only whether or not it is above zero. The variation in number of viable land uses over land units is portrayed in Figure 4. This diagram illustrates the range in the number of viable land uses across land types and regions.
~e~fo~~u~ume
22 Number 351991
REGIONS NO1
E03
CO2
CO3
WOZ
W03
SC04 SW04
SW05 SW06 SW07
Number of viable WCS
0
1-2
3-5
6.5
S-13
Figure 4. Range of viabie uses across land types and regions.
These data, by themselves, provide no logical basis fur adopting any specific number of viable land uses as a threshold for marginal land. The literature was searched for theoretical or empirical bases for specifying a threshold of viable uses below which agriculture becomes particularly risky.
given different specifications (~1 to ~5) of the number of viable uses. For ~~ustrative purposes we have adopted a threshold of three viable uses, Under this scenario Figure 5 provides a broad categorization af the proportion of marginal land for each census division (or part thereof).
Although no theoretical or empi~cally-delved threshold was apparent, research has established principles of risk and specialization, and the somewhat self-evident conclusion that, ceteris paribus, more risk exists where only one crop is viable than where several crops are viable (FOUND, 1971; SMLTK, 1986f. With a greater number of crops which can be profitably grown there is less economic dependence on one crop and hence less potential for failure. Of course, lands which can support only one or two viable uses are not to be confused with lands which have numerous viable uses but where ordy one or two uses are selected to achieve economies of specialization or scale. Lands with limited options are logically more vulnerable than those which can support a larger number of viable uses.
Marginal areas, defined on the basis of 25% or more of their farmland in the marginal category represent that part of Ontario beyond the agricultural heartland of the Southwest and Ontario Lakeshore. Even with this fairly liberal definition of marginal areas, the pattern indicated is not inconsistent with the regional distribution of agricultural characteristics often associated with marginal areas. Within this broad area, however, the proportion of land classified as marginal still ranges from 28 to 100%. The more detailed pattern within this zone accords well with recognized contrasts in the biophysical base for agriculture. The area with 100% of its farmland in the marginal category not only encompasses Northern Ontario but, despite the coarseness of the statistical units employed, demarcates quite effectively the shield area south of the French and Mattawa rivers, and captures the low heat unit Dundalk uplands, which constitute an outlier of climatic zone 2. To the east and southwest of the shield are two regions long
Table 4 provides info~atio~ on the proportion of the land cleared and available for agriculture in each census division which would be classified as marginal
Geoforum/Volume 22 Number 311991 Table 4. Percentage
343
of agricultural land classed as marginal: by census division and number of viable uses Number of viable uses
Number of viable uses Region
Census division
NO1
Algoma Cochrane Kenora Nipissing Manitoulin Rainy River Sudbury Thunder Bay Timiskaming
E03
co3
co2
wo2
I1 73 28 93 72 58 58 75 85 36
12
53
14
15
Region
100 100 100 100 100 100 100 100 100
100 100 100 100 100 100 100 100 100
100 100 100 100 100 100 loo’ 100 100
100 wo3 100 100 100 100 100 100 100 100 SC04
Dundas Frontenac Glengarry Grenville Lanark Leeds Lennox and Addington Ottawa-Carleton Prescott Renfrew Russell Stormont
39 51 50 48 52 77 77 39 58 18 42 31
39 51 50 48 52 77 77 39 58 18 42 31
65 72 82 70 73 87 89 66 88 71 73 72
65 72 82 70 73 87 89 66 88 71 73 72
Hastings (part) Muskoka (part Peterborough (part) Victoria (part)
42 33 22 24
42 33 22 24
57 100 37 41
57 100 37 41
57 100 37 41
Haliburton Hastings (part) Muskoka (part) Parry Sound Peterborough (part) Victoria (part)
100 100 100 93 100 100
100 100 100 93 100 100
100 100 100 100 100 100
100 100 100 100 100 100
100 100 100 SW06 100 100 100 SW07
26
26
100
100
100
Parts of Dufferin, Grey and Wellington
recognized as intermediate in position between the shield and the agricultural heartland. For the area commonly referred to as Eastern Ontario, lying between the St. Lawrence and Ottawa rivers the range of farmland in the marginal category is from 65 to 89%. From Prince Edward County jutting into Lake Ontario to the Bruce peninsula in Lake Huron we find a zone with 2!3-71% of its farmland in the marginal category. In both cases the major constraints stem from limitations imposed by soil quality. Conclusions
This paper has demonstrated the feasibility of defining the marginality of agricultural lands on the basis
65 72 82 70 73 87 89 SW04 66 88 71 73 72 SW05
Census division
51
12
53
14
Bruce Dufferin (part) Grey (part) Perth Simcoe Waterloo Wellington (part)
23 14 25 12 18 6 7
28 29 28 13 30 14 13
71 44 47 13 43 22 18
71 71 44 44 47 47 13 13 43 43 22 22 18 18
Durham Halton Northumberland Peel Prince Edward York Metro Toronto
9
9
9
9
9
15 17 7 28 14 7
15 17 7 28 14 7
15 17 7 28 14 7
15 17 7 28 14 7
15 17 7 28 14 7
6 7 2 1
6 7 2 1
6 7 2 1
6 7 2 1
20 10 3 6
Elgin Haldimand-Norfolk Hamilton-Wentworth (part) Kent (part) Lambton Middlesex (part) Niagara (part)
15 20 9
15 20 9
15 20 9
15 20 9
15 20 9
8 2 6 20
8 2 6 20
8 2 6 20
8 2 6 20
8 2 6 20
Essex Kent (part)
19 14
19 14
19 14
19 14
19 14
Hamilton-Wentworth (part) Niagara (part)
14
14
14
14
14
Brant Huron Middlesex (part) Oxford
55
of the theoretical viability of uses, which in turn can be specified as a function of inherent biophysical characteristics of lands. The approach is founded on principles relating to the economics of land uses. The procedures allow inclusion of variability in biophysical and economic conditions by incorporating probabilities in the definitions of viable uses and marginal lands. The empirical demonstration for Ontario illustrated that this systematic approach is practicable. Furthermore, the results are intuitively appealing, generating areas of land marginality broadly consistent with conventional wisdom. Nonetheless, a number of noteworthy shortcomings are evident. The empirical
344
Geoforum/Volume
22 Number 3/1991
iwK
rllt Kh Oh-l
AKH
Figure 5. Percent of agricultural land’ with three or less viable uses by census division:* 1 = lands ‘cleared and available’ for agriculture, 2 = census divisions or parts thereof.
analysis was limited in that data were not available for all uses, and information on the distribution of lands with specified characteristics was not available in other than the aggregate form for census divisions. Finer spatial resolution would facilitate consideration of more sophisticated functions for defining the marginality of areas, for example by considering the local spatial extent and configuration of marginal lands. Also in our analysis uses are defined as crops, yet some crops, for example forages, can support a variety of agricultural enterprises and commodities, each of which may have distinct levels of economic viability. It is technically possible to use feed conversion ratios to translate crop productivities to livestockcarrying capacities (SMIT et al., 1984), and this exercise would be essential for an analysis of marginal farms rather than marginal lands. Nonetheless, the procedures as implemented in this paper are reasonable for crop-dominated systems, but further development in the definition of ‘uses’ and in the specifi-
cation of viability and marginal lands would be necessary in systems where multiple products or noncrop products are significant. Similar refinement would be necessary in systems where institutional conditions (e.g. quotas or supply management arrangements) greatly influence the economic viability of uses. For example, one could declare lands farmed without quota to have zero price for the products concerned, but this is hardly in keeping with the intent of the approach to define marginality relative to the stable characteristics of the land and the broad economic conditions applying to the system. These challenges regarding the treatment of price are symptomatic of comparative static (as opposed to general equilibrium) formulations. This paper considered alternative means of defining the marginality of lands based on the number and level of positive economic returns to uses. There are no established theories or principles with which to
Geoforum/VoIume
22 Number 3/1991
choose a threshold number of viable uses for marginality. Even if marginality of lands were to be defined according to mean positive economic returns, any threshold level for defining marginal lands would be arbitrary. On reflection, the predilection in the literature (this study included) to dichotomize marginality is probably counter-productive. Perhaps a more useful exercise would be to identify degrees of land marginality or viability. Certainly the framework developed in this paper lends itself to characterizing the marginality of both specific lands and broader regions as continuous variables. Pursuing this line of thinking further, there probably is merit in replacing the notion of marginal land with a concept of land versatility. Not only would this avoid potential confusion with the plethora of definitions of marginality but it may also have utility in evaluating sensitivities and prospects for agricultural regions. Although this paper’s principal focus is methodological, some practical applications of the analysis are apparent, particularly stemming from its potential to assess the viability or versatility of land areas independently of their current management. For example, for the targeting and design of agricultural or regional development initiatives it is important to separate those areas where the land imposes an inherent constraint on production from those where the land itself is not a significant limitation to viable agricultural use. Similarly, this ability to classify land according to its agricultural viability is relevant to the formulation and implementation of policies dealing with land retirement and conservation. The procedures also have implications for property taxation. Ontario has a differential assessment program whereby farmland is appraised and taxed on its agricultural value rather than on its market value, which is the general basis for property taxation. The rationale is to not penalize agricultural activities for the inflationary effects of non-farm demands on land. The analysis outlined in this paper provides a means of evaluating lands according to their agricultural potential, independent of both current management and non-farm demands. Acknowledgements-This
research was supported by the Ontario Ministry of Agriculture and Food and the Social Sciences and Humanities Research Council of Canada. The analysis was facilitated through the Land Evaluation Group at the University of Guelph. Suggestions provided by Julius
345
Mage, Deborah Bond, and anonymous reviewers are gratefully acknowledged.
References AElS (ATMOSPHERIC
ENVIRONMENT SERVICE) (1982) Canadian Climate Normals: Temperature and Precipitation, 1951-1980. Environment Canada, Ottawa. BAKER, 0. E. (1921) The increasing importance of the physical conditions in detemuning the utilization of land for agriculture and forest production in the United States, Ann. Ass. Am. Geogr., 11,17-46. BARLOWE,
R.
(1978)
Land
Resource
Economics.
Prentice-Hall, Englewood Cliffs, NJ. BEATTIE, K. G., BOND, W. G. and MANNING, E. W. (1981) The Agricultural Use of Marginal Lana? a Review and Bibliography. Lands Directorate, Environment Canada, Ottawa. BOND, D. M., HILTS, S. G., MCBRIDE, R. A. and MILLER, M. H. (Eds) (1981) A Demonstration of the Land Evaluation Project for Ontario Data Base. Technical Report No. 2/81, Land Evaluation Project, Centre for Resources Development, University of Guelph, Guelph. CHAMPION, A. G. (1983) Land use and competition, In: Progress in Rural Geography, pp. 2145, M. Pacione (Ed.). Croom Helm, London. COMEAU, J. E. (1974) Northern agriculture: northern Ontario and Quebec, Agrologist, 3(6), 18-22. DEPARTMENT OF TRANSPORT, METEOROLOGICAL BRANCH (1959-1976) Monthly Record of Meteorological Observations in Canada. Meteorological Branch, Department of Transport, Toronto. ENVIRONMENT CANADA (1977-1982) Monthly Record of Meteorological Observations in Eastern Canada,
Part 3, Vols 62-67, Nos 5-9. FOUND, W. C. (1971) A Theoretical Approach to Rural Land Use. Edward Arnold, London. FRANCIS, R. J. (1970) The significance and usage of the term marginal in fringe settlement studies, In: Geographic, pp. 23-40, H. D. Foster (Ed.). Western Geographic Series, 2, University of Victoria, Victoria. GRIGG, D. B. (1984) An Introduction to Agricultural Geography. Hutchinson, London. HALL, P. (1966) Von Thunen’s Isolated State. Pergamon Press, Oxford. HASLEIT, E. (1983) Yields Necessary to Attain Various Margin Levels with Alternative Product Prices. Unpublished report of the Regional Development Branch, Agriculture Canada. ILBERY, B. W. (1985) Agricultural Geography: a Social and Economic Analysis. Oxford University Press, New York. IRONSIDE, R. G., PROUDFOOT, V. B., SHANNON, E. N. and TRACIE, C. J. (1974) Frontier development and perspectives on the western Canadian frontier, In: Frontier Settlement, pp. l-45, R. G. Ironside et al. (Eds). Department of Geography, University of Alberta, Edmonton. KLAGES, K. H. W. (1949) Ecological Crop Geography. Macmillan, New York. LEG (LAND EVALUATION GROUP) (1983) Analysis
346
GeofondVolume
of the Production Possibilities of Ontario Agriculture: Regional Assessments of the Prospects for Sustainable Agricultural Production. Publication No. LEG-16,
University School for Rural Planning and Development, University of Guelph, Guelph. MANDALE, M. (1984) Marginal Land Utilization and Potential, Kent County, New Brunswick. Working Paper No. 31, Lands Directorate, Environment Canada, Ottawa. MANNING, E. W. and BARDECKI, M. J. (1988) The evaluation of resource management options: the case of marginal resource lands, Unpublished paper presented to the C.A.G., Halifax. MAXWELL, J. W. (1966) Notes on land use and landscape evaluation in a fringe area of the Canadian shield, Geogrl Bull., 8,134-150.
MCCARTY,
H. M. (1954) Agricultural
geography,
In:
American Geography, Znventory and Prospect, pp. 258-
277, P. E. James and C. F. Jones (Eds). Syracuse University Press, Syracuse, NY. McCUAIG, J. D. and MANNING, E. W. (1982) Agricul-
tural Land-use Change in Canada: Process and Consequences. Lands Directorate, Environment Canada,
Ottawa. OECD (ORGANIZATION FOR ECONOMIC COOPERATION AND DEVELOPMENT) (1964) Low Incomes for Agriculture. OECD, Paris. PACIONE, M. (Ed.) (1986) Progress in Agricultural Geography. Croom Helm, London. REEDS, L. G. (1964) Agricultural geography: progress and prospects, Can. Geogr., 8,51-63.
22 Number 3/1991
RICARDO , D . (1817) The Principles of Political Economy and Taxation. Dent, London. SCFI (SPECIAL COMMITTEE ON FARM INCOME) (1969) The Challenge of Abundance. Report of the Special Committee on Farm Income in Ontario, Toronto. SMIT, B., BRKLACICH, M., DUMANSKI, J., MacDONALD, K. B. and MILLER, M. H. (1984) Integral land evaluation and its application to policy, Can. J. Soil Sci., 64,467-479.
SMITH, W. (1986) Agricultural marketing and distribution, In: Progress in Agricultural Geography, pp. 219 238, M. Pacione (Ed.). Croom Helm, London. SYMONS, L. (1970) Agricultural Geography. Bell, London. TROUGHTON, M. J. (1977) Persistent problems of rural development in marginal areas of Canada, In: Rural Development in Highlands and High Latitude Zones, pp. w-107, L. Koutaniemi (Ed.). Proceedings of a Symposium held 22-28 August 1977, I.G.U. Commission on Rural Development, University of Oulu, Oulu. VARJO, U. (1979) Productivity and fluctuating limits of crop cultivation in Finland, Geographia Polo&a, 40, 225-233.
ZIMMER, B. E. and RODD, R. S. (1971) Socio-economic
Factors Related to the A.R.D.A. Programme for Consolidation and Farm Enlargement in Eastern Ontario.
Centre for Resources Development, University of Guelph, Guelph. ZIMMERMAN, E. W. (1951) World Resources and Zndustries. Harper, New York.