USF Libraries

The applicability of the river continuum concept to the upper reaches of a neotropical lower montane stream

MISSING IMAGE

Material Information

Title:
The applicability of the river continuum concept to the upper reaches of a neotropical lower montane stream
Translated Title:
La aplicabilidad del concepto del río continúo en la parte superior de una quebrada ( )
Physical Description:
Book
Language:
English
Creator:
Skaff, Nicholas
Publication Date:

Subjects

Subjects / Keywords:
River continuum concept   ( lcsh )
Invertebrates   ( lcsh )
Costa Rica--Puntarenas--Monteverde Zone   ( lcsh )
Concepto del río continuo
Invertebrados
Costa Rica--Puntarenas--Zona de Monteverde
Tropical Ecology Fall 2009
Ecología Tropical Otoño 2009
Genre:
Reports   ( lcsh )
Reports

Notes

Abstract:
The applicability of the River Continuum Concept (RCC) to the upper reaches of Quebrada Máquina, a lower montane stream in Monteverde, Costa Rica, was examined. Macroinvertebrate samples were taken from ten points along the stream from first through fourth order segments. The families of the collected individuals were then categorized based on functional feeding group. The similarity between the families found at each collection point was calculated, along with correlations between the functional groups and various stream characteristics. In most cases, RCC predictions did not apply to Quebrada Máquina. The first (first order) and last (fourth order) sample points were 92% similar despite RCC predictions of substantial divergence in relative functional group abundance. This digression from the RCC predictions may be caused by the relatively few differences in stream characteristics between first and fourth order sections. Specifically, the observed similarities and correlations may have been determined by local scale heterogeneity of the stream characteristics.
Abstract:
Se estudió la aplicabilidad del Concepto Río Continuo (CRC) en la parte superior de la Quebrada Máquina, una quebrada de montaña baja, en Monteverde, Costa Rica. Las muestras de Macroinvertebrados fueron reunidas de diez puntos en la quebrada en partes del orden primero al cuarto. Las familias de los individuos reunidos fueron clasificadas por grupos de alimentación funcional. La similitud entre las familias encontradas en los puntos fue calculado, junto con las correlaciones entre los grupos funcionales y las características de la quebrada. Muchas veces, las predicciones del CRC no aplican a la Quebrada Máquina. El primero (orden primero) y último (orden cuarto) puntos eran 92% similares, a pesar de las predicciones del CRC con diferencias grandes en la abundancia de los grupos funcionales. La digresión en las predicciones de CRC puede ser causada por las pocas diferencias en las características de la quebrada en las partes del orden primero y el cuarto. Específicamente, las semejanzas y correlaciones observadas podían ser determinadas por heterogeneidad local en las características de la quebrada.
Language:
Text in English.
General Note:
Born Digital

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - M39-00331
usfldc handle - m39.331
System ID:
SFS0001500:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xlink http:www.w3.org1999xlink xmlns:xsi http:www.w3.org2001XMLSchema-instance
leader 00000nas 2200000Ka 4500
controlfield tag 008 000000c19749999pautr p s 0 0eng d
datafield ind1 8 ind2 024
subfield code a M39-00331
040
FHM
0 041
eng
049
FHmm
1 100
Skaff, Nicholas
242
La aplicabilidad del concepto del ro contino en la parte superior de una quebrada
245
The applicability of the river continuum concept to the upper reaches of a neotropical lower montane stream
260
c 2009-11
500
Born Digital
3 520
The applicability of the River Continuum Concept (RCC) to the upper reaches of Quebrada Mquina, a lower montane stream in Monteverde, Costa Rica, was examined. Macroinvertebrate samples were taken from ten
points along the stream from first through fourth order segments. The families of the collected individuals were then categorized based on functional feeding group. The similarity between the families found at each collection point was calculated, along with correlations between the functional groups and various stream characteristics. In most cases, RCC predictions did not apply to Quebrada Mquina. The first (first order) and last (fourth order) sample points were 92% similar despite RCC predictions of substantial divergence in relative functional group abundance. This digression from the RCC predictions may be caused by the relatively few differences in stream characteristics between first and fourth order sections. Specifically, the observed similarities and correlations may have been determined by local scale heterogeneity of the stream characteristics.
Se estudi la aplicabilidad del Concepto Ro Continuo (CRC) en la parte superior de la Quebrada Mquina, una quebrada de montaa baja, en Monteverde, Costa Rica. Las muestras de Macroinvertebrados fueron reunidas de diez puntos en la quebrada en partes del orden primero al cuarto. Las familias de los individuos reunidos fueron clasificadas por grupos de alimentacin funcional. La similitud entre las familias encontradas en los puntos fue calculado, junto con las correlaciones entre los grupos funcionales y las caractersticas de la quebrada. Muchas veces, las predicciones del CRC no aplican a la Quebrada Mquina. El primero (orden primero) y ltimo (orden cuarto) puntos eran 92% similares, a pesar de las predicciones del CRC con diferencias grandes en la abundancia de los grupos funcionales. La digresin en las predicciones de CRC puede ser causada por las pocas diferencias en las caractersticas de la quebrada en las partes del orden primero y el cuarto. Especficamente, las semejanzas y correlaciones observadas podan ser determinadas por heterogeneidad local en las caractersticas de la quebrada.
546
Text in English.
650
River continuum concept
Invertebrates
Costa Rica--Puntarenas--Monteverde Zone
4
Concepto del ro continuo
Invertebrados
Costa Rica--Puntarenas--Zona de Monteverde
653
Tropical Ecology Fall 2009
Ecologa Tropical Otoo 2009
655
Reports
720
CIEE
773
t Monteverde Institute : Tropical Ecology
856
u http://digital.lib.usf.edu/?m39.331



PAGE 1

The Applicability of the River Continuum Concept to the Upper Reaches of a Neotropical Lower Montane Stream Nicholas Skaff Department of Biology, Tufts University ABSTRACT The applicability of the River Continuum Concept (RCC) to the upper reaches of Quebrada M‡quina, a lower montane stream in Monteverde, Costa Rica, was examined. Macroinvertebrate samples were taken from ten points along the stream from first through fourth order segments. The families of the collected individuals were then categori zed based on functional feeding group. The similarity between the families found at each collection point was calculated, along with correlations between the functional groups and various stream characteristics. In most cases, RCC predictions did not app ly to Quebrada M‡quina. The first (first order) and last (fourth order) sample point s were 92% similar despite RCC predictions of substantial divergence in relative functional group abundance. This digression from the RCC predictions may be caused by the relatively few differences in stream characteristics between first and fourth order sections Specifically, the observed similarities and correlations may have been determined by local scale heterogeneity of the stream characteristics. RESUMEN Se estudi o la aplicabilidad del Concepto R’o Continuo (CRC) a la parte superior de Quebrada M‡quina, una quebrada de monta–a baja, en Monteverde, Costa Rica. Las muestras de m acroinvertebradas fueron reunida s de diez puntos en la quebrada en partes de orden primer o a cuarto. Las famil ias de los individuos reunidos fueron clasificada s por grupos de alimentaci—n funcional. La similitud entre las familias encontradas en los pu ntos fueron calculados, junto con las correlaci—nes entre los grupos funciones y las caract er’sticas de la quebrada. Muchas veces, las predicciones de CRC no aplica n a Quebrada M‡q uina. El primero (orden primero) y œ ltimo (orden cuarto) puntos eran 92% similar es a pesar de las predicciones de CRC con di ferencias gra ndes en la abundancia de lo s gru pos funcionales. La digresi—n en las predicciones de CRC puede ser causada por las pocas differencias en las caracter’sticas de la quebrada en las partes de orden primero y cuarto. Espec’ficame nte, las semejanzas y correlacio nes o bservadas pod’an se r determinadas por heterogeneidad local en las caracter’sticas de la quebrada. INTRODUCTION Dynamic abiotic and biotic interactions in pristine river ecosystems are extremely consistent due to the predictable biological responses to the physical componen ts (Vannote et al. 1980). The River Continuum Concept (RCC) endeavors to systematically illustrate these patterns on a comprehensive scale. The basis of the RCC rests on the view that throughout the longitudinal river gradient, biological communities dyna mically equilibrate with abiotic factors. In other words, the biological organization is dependent on the specific local allocation of kinetic energy through the physical system. For example, headwater community

PAGE 2

structure is largely shaped by riparian ve getation, which adds organic detritus to the channel and shades the river bottom, limiting autotrophic production. The specific energy distribution (including the influence of riparian vegetation) is greatly influenced by river size, which makes size cate gorization very important to the RCC (Vannote et al. 1980). The smallest consistently flowing stream is referred to as first order. Its mergence with another first order stream forms a second order and the coalescence of two second order streams produces a third order and so on (Allan et al. 2007). The RCC generally accounts for river orders ranging from one to twelve (Thorp & Covich 1991). Aquatic macroinvertebrates are often utilized in studying the biological responses of the system because they clear ly reflect changes in food resource availability in relation to stream size (Vannote et al. 1980). Different abundances of specific functional feeding groups can be found along the longitudinal gradient. In general, many downstream organisms exploit upst ream inefficiency in energy capture. Shredding (microbes gleaned from course organic matter like leaves) and collecting (microbes collected from direct water transport or fine sediment/organic matter) characterize first and second order stream macroinvert ebrate species, while collecting and grazing (algae removed from substrate) are most commonly seen in mid size fourth or fifth order rivers (Vannote et al. 1980). Near the mouth, collectors dominate, while predator abundance stays consistent throughout th e system (Vannote et al. 1980; Hawkins & Sedell 1981). Despite its temperate origins, the RCC has been applied successfully in the tropics (Greathouse & Pringle 2006). However, there may be some circumstances in which trends in macroinvertebrate functio nal group structure differ from those expected by the continuum theory. Macroinvertebrates may persist in environments that the RCC would predict to be unsuitable by capitalizing on extreme local heterogeneity in the stream characteristics (Lancaster & Be lyea 1997). Nevertheless, I hypothesized that, overall, general trends in stream structure w ould dominate and macroinvertebrate species composition w ould follow RCC predictions in Quebrada M‡quina, a neotropical lower montane stream. METHODS Study Loca tion Macroinvertebrates were collected at ten points along Quebrada M‡quina in Monteverde, Costa Rica between 1690 m (Sample Site 1) and 1440 m (Sample Site 10). The stream originates high in the Tilar ‡ n Mountain Range (maximum elevat ion 1850 m), which r epresents the continental divide (Jankowski & Rabenold 2007). Beginning above the Estaci—n Biol—gica in lower montane forest, the stream extends into the adjacent valley. On two separate occasions, five samples were taken every 30 m increase in elevation above the Estaci—n and every 15 m drop in elevation after the waterfall below the Estaci—n. The elevation adjustment was made to accommodate a decrease in stream slope in the lower half. In each of the two collection round s one sample was taken in both first and second order sections of the stream, three samples were taken in third order sections, and five samples were collected in fourth order sections. The sample areas were rectangles five meters long (parallel with stream flow) and the width of the stream (perpendicular to stream flow). This site size was selected to maximize macroinvertebrate capture and limit local site variation like elevation and substrate composition.

PAGE 3

Sampling Sampling occurred between 9 November and 20 November 2009. Prior to macroinvertebrate collection, local stream characteristics were noted. The approximate distance from the nearest upstream waterfall, the percent substrate composition of rock, sand and vegetation detritus, stream width, depth, grade and elevation were all recorded. Macroinvertebrates were collected for 45 minutes at each sample point using a benthic macroinvertebrate collection sieve. The substrate was disturbed for approximately 45 seconds using a hand rake, while the sieve was held downstream. Disl odged invertebrates in the sieve were then preserved in vials containing 80% ethanol. This process was repeated many times over the sample period in the entire range of local stream bottom compositions (vegetation detritus, rocks, sand). Identification Macroinvertebrate samples were identified in the lab using a dissecting microscope. The identifications were made with Lehmkuhl (1979) as a reference. Individuals were identified to family with the exception of leaches, Stylommatophora, and Isopoda (excl uding Armadillidiidae). In order to assess the applicability of the River Continuum Concept, individual macroinvertebrate families from specific sample sites were grouped into their respective function feeding groups using Merritt and Cummins (1996). Dat a Analysis The relative abundance of each functional feeding group was calculated by site. Morisita's Index of Similarity was used to measure the percent similarity between macroinvertebrate samples at each site using families as taxa. With JMP 5.0.1a, the similarity percentages were combined to form a hierarchical cluster diagram, grouping the sites with the most similar macroinvertebrate assemblages (Krebs 1999). Pairwise correlations were calculated between functional feeding group and site location, stream characteristics and stream order. Finally, elevation and the local stream characteristics were tested for correlation RESULTS Overall, 572 individual macroinvertebrates were collected, representing 12 orders and 24 families (Appendix A). The most similar sites based on macroinvertebrate family and functional feeding group seem to be unrelated spatially (Figure 1). For example, Sites 1 and 10 are approximately 92% similar in macroinvertebrate family abundance and presence, yet they represent c omplete opposite ends of the longitudinal spectrum (Table 1). When the families are placed in function groups, the same site similarities are apparent (Figure 1, for specific abundances see Appendix B). Further confirmation can be seen in the lack of sig nificant correlation between relative functional group abundance and site location (Table 2). However, some of the recorded stream characteristics (Table 4) show significant correlations (Table 3) with specific functional groups. There is a negative rela tionship between the proportion of grazers and the sandiness of the substrate. A higher percentage of

PAGE 4

sand on the stream bottom indicates a lower proportion of grazers. Nevertheless, there is no significant correlation between the substrate sandiness and elevation (Table 5) Also, grazer abundances are negatively correlated with water depth and shredder abundances are positively correlated with stream width. All other stream characteristic and functional group relationships are not statistically signifi cant. Despite the relationships between stream characteristics and functional group abundance, there are no significant relationships between functional group and site location (Table 2) The strongest is a marginally significant negative relationship be tween site and predator abundance. Conversely, quite a few significant correlations can be found between elevation and other recorded stream characteristics (Table 5) Elevation is positively correlated with vegetation detritus and negatively correlated with both water depth and stream width. This means that higher order streams (lower elevations) tend to have substrates with less vegetation detritus and that lower order streams are both wider and deeper than those found at higher elevations. DISCUS SION It appears that the stage has been properly set for the River Continuum Concept's macroinvertebrate predictions in that the appropriate abiotic and biotic stream characteristics are correlated. It is predicted that down an elevational or stream ord er gradient there will be stream widening and deepening, which has been found. Also supported by the results, the concept predicts that with stream widening there will be less vegetation detritus in the substrate (Vannote et al. 1980). However, the trend s found in relative functional group abundance do not correspond with tendencies predicted by the RCC. Following a review of the various discrepancies, an explanation will be provided. The strong negative correlation found between predator abundance and stream order is in direct opposition to what is expected by the RCC. The RCC assumes that the other three functional groups depend on one general type of food source (e.g. Grazers depend on algae etc.). Fluctuations in this resource cause the proportion s of these functional groups to change. However, predator populations only depend on the relative abundances of the other functional groups. Therefore, no matter how the grazers, shredders and collectors respond to resource availability, the relative pre dator population should stay the same (Vannote et al. 1980). For this reason, the observed change in the proportion of predators is quite surprising. The actual proportions of grazers also run contrary to the expected. The RCC assumes that the highest p roportion of grazers will be present in mid order streams due to the elevated levels of algal growth associated with a wider stream and a relatively shallow channel (Vannote et al 1980). The data from this study illustrates that this is not the case. Th e lack of correlation between site number and shredder abundance is also unexpected. Shredders are typically higher in abundance in the headwaters due to more leaf detritus off of which they feed. Finally, according to the RCC no large correlation shou ld be found between site and collector abundance in low to mid order streams. The correlations did support this prediction (there was no correlation), however, this was not due to consistency in the observed proportion of collectors, but rather the lack o f consistency High proportions were found intermixed with low proportions at various points throughout the stream order gradient (Figure 1). According to the RCC, relative collector abundances should stay around 50% in the sampled portions of Quebrada M ‡quina (Vannote et al. 1980).

PAGE 5

One possible explanation for the disparity in observed and expected feeding group proportions may be the relatively small ranges of physical stream characteristics found between the stream orders. For example, the first orde r and fourth order portions of Quebrada M‡quina only differed in width by 2.5 m. However, just third and fourth order sections of the Dolores River in Colorado ranged from 7.3 m to 16.8 m wide over an elevation range of 500 m (Canton & Chadwick 1983). Th e relatively small width change in Quebrada M‡quina may limit the disparity between orders, making any differences in functional group presence much less apparent. The use of order to characterize a stream is sometimes deemed to be a misrepresentation of the true size. Some consider drainage area and discharge to be better indicators (Allan 2007). If the drainage area and discharge rates of Quebrada M‡quina equate best with the typical first order stream, then it may be best to expect first order macroin vertebrate abundances. The RCC's failure to explain the observed macroinvertebrate presence may be more a failure of the stream order classification system than the RCC itself. Future research could investigate this relationship. In a stream that shows little size change across orders, the specific results found may best be explained by the local heterogeneity of stream characteristics. The results from Quebrada M‡quina seem to illustrate this idea. Of the correlations performed, the highest number of s ignificant correlations was found in comparisons of local stream characteristics and functional group abundance. For example, the relative proportion of grazers was significantly correlated with the water depth and the sandiness of the substrate, but was not correlated with stream order. Randomly, some of the sample sites may have contained deeper water or a sandier substrate. A tree fall across the stream may have raised water depth and caused sediment build up. An event such as this is not influenced by stream order, but rather by chance. Incidences like this one may frequently be masked by the more general trends in water depth and substrate composition such as those predicted by the RCC. However, the narrow stream characteristic ranges in Quebrada M‡quina possibly accentuated chance heterogeneity and precluded correct RCC predictions. This is supported by DeAngelis and Waterhouse's (1981) conclusion that the idea of stream equilibrium (essentially the RCC) depends greatly on the scale of the study. On a large scale, the overall effect may be greater than the individual components, enabling the function of systems like the RCC (Allen & Hoekstra 1992). ACKNOWLEDGEMENTS I would like to thank Pablo Allen for his courage and fortitude in helping me pl an this project and for the many hours he spent helping me identify the macroinvertebrate families. Also, I would like to express appreciation for Anna Stuart's tireless help in teaching me stream sampling and water testing techniques. I would like to gi ve a small but special thanks to Blaine Marchant for literally getting me on the right path. Finally, I want to thank CIEE: Costa Rica and the Estaci—n Biologica for giving me the opportunity to gain valuable experience in field research in one of the wor ld's most beautiful forests.

PAGE 6

LITERATURE CITED Allan, J.D., and M.M. Castillo. 2007. Stream Ecology: Structure and Function of Running Water. 2 nd Ed. Springer, Netherlands. Allen, T.F.H, and T.W. Hoekstra. 1992. Toward a unified ecology. Columbia Un iversity Press, New York. Canton, S.P. and J.W. Chadwick. 1983. Season and longitudinal changes in macroinvertebrate functional groups in the Dolores River, Colorado. North American Benthological Society 2 (1): 41 47. DeAngelis, D.L., and J.C. Waterhous e. 1987. Equilibrium and nonequilibrium concepts in ecological models. Ecological Monographs 57: 1 21. Greathouse, E.A., and C.M. Pringle. 2006. Does the river continuum concept apply on a tropical island? Longitudinal variation in a Puerto Rican stream. Can. J. Fish Aquat. Sci 63: 134 152. Hawkins, C.P. and J.R. Sedell. 1981. Longitudinal and seasonal changes in functional organization of macroinvertebrate. Ecology 64 (2): 387 397. Jankowski, J.E. and K.N. Rabenold. 2007. Endemism and local rarity in birds of neotropical montane rainforest. Biological Conservation 138: 453 463. Krebs, C.B. 1999. Ecological Methodology. Addison Wesley Longman, Menlo Park. Lancaster, J. and L.R. Belya. 1997. Nested hierarchies and scale dependence of mechanisms of fl ow refugium use. Journal of the North American Benthological Society 16 (1): 221 238. Lehmkuhl, D.M. 1979. How to Know the Aquatic Insects. Pictured Key Nature Series, Dubuque, Iowa. Merritt, R.W., and K.W. Cummins 1996. An Introduction to the Aquatic I nsects of North America. Kendall/Hunt, Dubuque, Iowa. Thorp, J.H. and A.P. Covich, eds. 1991. Ecology and Classification of North American Freshwater Invertebrates. Academic Press, New York.

PAGE 7

Figure 1: Site Similarity Clusters of Sampled Macroinvertebrates. Sa mple si tes in Quebrada M‡quina, Monteverde, Costa Rica are grouped based on the similarity of the macroinvertebrate families present. Groups closer to the left axis are the most similar. The pie charts represent the relative abundance of each macroinvertebrate functional group at each site (numerical data in Apdx. B).

PAGE 8

Table 3: Pairwise Correlations Between Reco rded Stream Characteristics and Macroinvertebrate Functional Feeding Groups in Quebrada M‡quina, Monteverde, Costa Rica. Site Number 1 2 3 4 5 6 7 8 9 10 1 100% 2 47% 100% 3 50% 53% 100% 4 91% 54% 51% 100% 5 61% 52% 42% 82% 100% 6 85% 40% 44% 87% 72% 100% 7 55% 13% 24% 66% 69% 67% 100% 8 88% 26% 40% 85% 54% 86% 71% 100% 9 84% 28% 41% 91% 78% 88% 83% 93% 100% 10 92% 28% 41% 86% 51% 82% 63% 96% 88% 100% Variable A Variable B Rho N Value P Value Site Collector 0.41 10 0.24 Grazer 0.56 10 0.09 Predator 0.61 10 0.0 6 Shredder 0.4 0 10 0.35 Variable A Variable B R ho N Value P Value % Sand Collector 0.03 10 0.92 Grazer 0.71 10 0.02 Predator 0.29 10 0.41 Shredder 0.5 0 10 0.14 Stream Width Collector 0.61 10 0.06 Grazer 0.44 10 0.2 0 Predator 0.42 10 0.23 Shredder 0.64 10 0.05 Strea m Depth Collector 0.09 0.8 Grazer 0.75 10 0.01 Predator 0.54 10 0.1 Shredder 0.31 10 0.37 Table 1: Percent Similarity (Morisita's Index of Similarity) of Macroinvertebrate Families Between Sample Sites. Sites one through ten represent adjacent sample sites in Quebrada M‡quina, Monteve rde, Costa Rica with site one in the first order stream section (highest elevation) and site ten in the lowest fourth order section. Table 2: Pairwise Correlations Between Sample Site and Macroinvertebrate Function Feeding Group in Quebrada M‡quina, Mon teverde, Costa Rica.

PAGE 9

Table 4: Recorded Stream Characteristics at Sample Sites in Quebrada M‡quina, Monteverde, Costa Rica. Distance from waterfall represents the horizontal distance from the sample site to the nearest upstream waterfall. The percent rock, vegetation detritus, and sand represent the total substrate composition at each sample site. The percent grade is the slope of the streambed. Distance Substrate Site from Waterf all % Rock % Veg Detritus % Sand Water Depth Width % Grade Elevation 1 50 60 40 0 2 1 10 1690 2 3 40 60 0 4 0.75 35 1660 3 25 90 10 0 8 1.5 10 1630 4 35 70 30 0 5 2 20 1600 5 50 80 20 0 8 2 10 1570 6 10 8 0 20 0 11 2 10 1500 7 35 35 5 60 22 3 5 1485 8 50 85 5 10 11 3.5 15 1470 9 50 75 5 20 21 3.5 10 1455 10 50 80 10 10 20 2 20 1440 Table 5: Pairwise Correlations Between Sample Site Elevation and the Recorded Stream Characteristics in Quebrada M‡quin a, Monteverde, Costa Rica. Variable A Variable B Rho N Value P Value Elevation Distance WC 0.34 10 0.03 % Rock 0.25 10 0.48 % Veg. Det. 0.77 10 0.008 % Sand 0.49 10 0.14 Water Depth 0.87 10 0.0009 Width 0.82 10 0.003 % Grade 0.29 10 0 .41

PAGE 10

Appendix A: Complete Macroinvertebrate Sample Data for Quebrada M‡quina, Monteverde, Costa Rica. Total abundance for families found at each site are listed in the far right column.

PAGE 11

Appendix B: Relative Abundance of Macroinvertebrate Functional Groups in Site Samples Taken at Quebrada M‡quina, Monteverde, Costa Rica. Site Collector Grazer Predator Shredder 1 37.3% 11.9% 22.4% 28.4% 2 15.4% 11.5% 42.3% 30.8% 3 20.8% 33.3% 20.8% 25.0% 4 37.2% 14.0% 20.9% 27.9% 5 16.2% 9.5% 20.3% 54.1% 6 31.4% 13.7% 21.6% 33.3% 7 13.5% 1.9% 13.5% 71.2% 8 56.0% 4.8% 3.2% 36.0% 9 47.1% 1.5% 13.2% 38.2% 10 56.3% 9.4% 6.3% 28.1%