Development and Use of a Percent Model Affinity for Assessment of Puerto Rico Streams
Report prepared by:
*V*V'5^
James Kurtenbach, Aquatic Biologist
Monitoring Operations Section
Approved by:
JoMnfKushwara, Chief
Monitoring and Assessment Branch

-------
TABLE OF CONTENTS
List of Figures	ii
List of Tables	iii
List of Acronyms	iv
Executive Summary	v
Introduction	1
Methods	1
Results and Discussion	6
Literature Cited	9
i

-------
LIST OF FIGURES
1.	Map of stream sites in Puerto Rico	2
2.	Map of stream sites in Puerto Rico. Green squares: Sampled in both 2013 and 2009. Blue
circles: Sampled in 2013only. Red diamonds: Sampled in 2009 only	3
3.	PMA score thresholds for biological condition categories (good, fair, poor)	5
4.	Box and whisker plots of the PMA and Mil for two different levels of forest cover	7
5.	Mil and PMA scores for the Rio Piedras sites	7
6.	Change in PMA scores at six Puerto Rico stream sites during the study period	8
li

-------
LIST OF TABLES
1. Example calculation of a PMA score	5
111

-------
LIST OF ACRONYMS
CV	Coefficient of Variation
EPA	Environmental Protection Agency
EPT	Ephemeroptera, Plecoptera, Trichoptera
HBI	Hillsenhoff Biotic Index
Mil	Macroinvertebrate Integrity Index
PMA Percent Model Affinity
iv

-------
EXECUTIVE SUMMARY
Puerto Rico currently lacks a stream monitoring program with direct assessment and reporting on
biological conditions and protocols for the biological assessment of Caribbean streams are
generally unavailable. Measures of macroinvertebrate community composition including percent
model affinity (PMA) have been developed and successfully applied in temperate North
America. The PMA was originally developed for use on New York State streams and measures
the similarity of a collected sample to that of an expected macroinvertebrate sample collected
from riffle habitat at minimally disturbed sites. High percent similarity is indicative of least
disturbed conditions and low similarity generally reflects poor habitat and/or water quality.
Original authors of the PMA suggested it could potentially be applied in other geographical
regions with some modifications. This report describes the results of the development and
testing of the PMA for its effectiveness on wadeable streams with riffle habitat located in Puerto
Rico.
Data used in the development and testing of the PMA were collected from sampling conducted
over the survey years (1994, 2006, 2009, 2011, 2013), at a total of 260 stream sites located in the
eastern, central and western mountain drainages of Puerto Rico. All stream sites varied in their
level of disturbance based on specific landscape features and human activities.
A set of 42 reference sites were used to establish a model community and is represented by eight
major organism groups. Mean average percent of each major aquatic group at these sites were
trichoptera (16%), lepidoptera (2%), ephemeroptera (31%), coleoptera (18%), diptera (22%),
odonata (4%), Crustacea (4%), and other (2%).
The proposed PMA was tested for its accuracy and variability (sampling and seasonal) with
independent data collected over the period 1994 to 2013. Accuracy of the PMA was evaluated
against two land use gradients and compared to the response of a macroinvertebrate integrity
index (Mil) already developed for use in Puerto Rico. The first evaluation assessed a group of
reference quality versus highly deforested stream sites. The PMA did not distinguish this
gradient with the same degree of separation as did the Mil, however, the PMA response was still
moderately sensitive. The second land use gradient assessed was in a small urban watershed.
Both the PMA and Mil tracked change of declining water quality from moderately developed
headwaters to the highly urban downstream reach. Sampling and seasonal variability of the
index was examined. Sampling and seasonal variability was similar or lower when compared to
other common community measures and indices used in the determination of water quality.
Results of this study will be used to make recommendations to the Puerto Rico Environmental
Quality Board (PREQB) for their stream monitoring and assessment program. Advantages of the
application of the PMA in Puerto Rico include, ease in which biological conditions can be
assessed without the need for extensive information and training on invertebrate taxonomy, and
the index is relatively easy to compute as compared to other indices.
v

-------
Introduction
Puerto Rico currently lacks a stream monitoring program with direct assessment, and reporting
on biological conditions and protocols for the biological assessment of Caribbean streams are
generally not available. Measures of macroinvertebrate community composition, including
percent model affinity (PMA), have been developed and successfully applied in temperate North
America. The PMA developed for shallow New York State streams (Novak and Bode 1992),
measures the similarity of a collected sample to that of an expected macroinvertebrate sample
collected from riffle habitat at minimally disturbed sites. Conceptually the model is developed
for an expected community and affinity to the model is measured using a percent similarity
index. High percent similarity is indicative of least disturbed conditions and low similarity
represents severe water quality degradation. Authors of the PMA suggested it could potentially
be applied in other geographical regions with some modifications.
Here the model was tested for its effectiveness on wadeable streams with riffle habitat located in
Puerto Rico. The PMA developed and tested for Puerto Rico streams is meant to complement
other community measures and provide additional water quality information not contributed by
other measures.
Methods
Study area
All benthic macroinvertebrate sampling was conducted in March-April over the survey years
(1994, 2006, 2009, 2011, 2013), at a total of 260 stream sites located in the eastern, central and
western mountain drainages of Puerto Rico (Figures 1 and 2). The area of study included three
geographic regions, humid east central, rainy west central, and rainy Luquillo Mountains, all
characterized by relatively heavy rainfall and steep topography. Stream size ranged from 1st to
4th order. Stream sites were exclusively high and moderate in gradient and dominated by riffle-
run habitat. The bottom substrate at all sites was similar, consisting mostly of small boulder and
cobble with lesser amounts of gravel, sand, and silt. The catchment area upstream of each site
had varied land use.
1

-------
1994, 2006, 2009, 2011
Puerto Rico Biological Stream Survey's
• m	«	^	«
„ . • %	V V*	• •• ,»7 >
• • •	•	• • ® V • - * 9	*1
• «•	#•	_ . * • • ®* #
• ••	I	. »v • . .
a * *	%S	¦ »«§>•& «**
a u >.	«•	« ® • a,
\	* . , 1 mt
m
® •
SEPA A
Stream Sites
*	1994 Targeted n = 105
¦	2006 Probabilistic Reference n = 19
N
Kilometers	•	2009 Probabilistic n = 50
50
I *	2011 Probabilistic Reference n = 20
Figure 1. Map of stream sites in Puerto Rico.
2

-------
CamuQ
- iMfla
snMniiJIn
vagt Alia
VITT lljWC'OW
t"Tj L'f 'j^jTC
CnboRaio ,„GniMn Salvia
^nwv-

Cilfrn II;
Wilift'f'' ffrftmc
i«^, Sao Juan
G3 ftid PiwW.
n J*.	Carolina . H«o«rerik^ ,
«	| Injj'tw W»	«5r &
r.uT>«iii . i h=m	-y un)	O
¦	^	t> YlyflOlA?®
¦	-^Surrfag.
cagfaw 7^5
J	i.	j^WflWiOh''
¦ ¦ ¦
Y«£WCQSI -» 1
HOlJllBD	HSU •
Cu'frj-afT^ .
s>
Figure 2. Map of stream sites in Puerto Rico. Green squares: Sampled in both 2013 and 2009; Blue
circles: Sampled in 2013 ONLY; Red Diamonds: Sampled in 2009 ONLY.
Site selection
Sites selected for the 1994 stream survey were easily accessed from road bridge crossings, while
the total number of sites sampled was proportional to the size of the drainage area. Sampling
locations selected for the 2006, 2009, 2011, and 2013 surveys used a probabilistic survey design
(Olsen et al. 1999). Results from these surveys were used to develop and test the PMA index.
Site selection did not include weighting in the number of stream sites from 1st through 4th order
stream classes. A set of reserve sites was chosen and served as replacements when target sites
were dry, inaccessible due to physical barriers or access was denied by the landowner.
Macroinvertebrate collection and sample processing
At each stream site a single riffle area was chosen for the collection of a benthic
macroinvertebrate sample. Benthic macroinvertebrates were captured from rock substrates
(cobble and small boulder) using a large frame rectangular kick net, constructed with an 800 x
900 mil mesh net (Bode 1991). Sample collection was accomplished by placing the kick net on
the stream bottom, mid-river in the riffle, while gradually working the net downstream and
laterally (approximately two net widths) for 5 minutes. Bottom substrate was vigorously
disturbed by foot so that sufficient organisms would be dislodged and swept into the net.
After the sample was collected, large debris were removed from the net, inspected for organisms,
and discarded. Macroinvertebrates and debris retained in the kick net were removed and placed
in a one liter container and preserved with 10% buffered formalin. Sample containers were
labeled with the appropriate site information and delivered to the EPA Edison laboratory. Kick
nets were thoroughly rinsed and cleaned between stream sites. At the laboratory a discrete 100-
organism subsample was sorted out and individuals were identified to the genus level of
taxonomy. Midges and worms were only identified to family level.

-------
Index development
PMA applied here is a measure of aquatic invertebrate composition among organism groups that
comprise the macroinvertebrate community of Puerto Rico streams. These major groups are
Trichoptera, Lepidoptera, Ephemeroptera, Coleoptera, Diptera, Odonata, Crustacea, and other.
The "other" category includes all other invertebrates not fitting in the above groups, such things
as worms and snails. Two hundred and sixty sites were used in the development of the model
that spanned a range of biological conditions as measured with a macroinvertebrate integrity
index (Mil), which scores sites in one of three biological condition categories (good, fair, poor).
All 260 sites varied with minimally to highly disturbed conditions as the result of various
landscape and site specific human disturbance. A set 42 reference sites were used to establish
the model community. Mean average percent of each major aquatic group at these sites was as
follows: Trichoptera 16.4%; Lepidoptera 1.75%, Ephemeroptera 31.10%; Coleoptera 17.89%;
Diptera 22.35%; Odonata 3.72%, Crustacea 4.37%; Other 2.25%. All reference site PMA scores
had a coefficient of variability that was approximately 17%. In general, the Trichoptera,
Lepidoptera, Ephemeroptera, and Coleoptera represent groups that are intolerant of poor water
quality, with Odonata and Crustacea having intermediate tolerance, while Diptera and other
groups are considered more tolerant of pollution. These should be taken as broad generalizations
since information on Puerto Rico invertebrate pollution tolerances is limited. Percent model
affinity is usually determined by shifts in the dominance of intolerant or tolerant groups. As
intolerant groups become less dominant and tolerant groups more dominant, affinity to the model
decreases and reflects an increase of water pollution and/or habitat degradation. On occasion a
stream site may have a dominance of one or two less tolerant groups which causes a departure
from the model community and is not related to pollution. In such circumstances PMA scores
should be carefully interpreted and combined with information from other measures. For
example, some Puerto Rico stream sites have a high abundance of leptophlebiid mayflies.
Affinity to the model community is determined using a percent similarity index (Novak and
Bode 1992). The percentages of the major groups in the sample and should equal 100. Then the
lessor of the two values (model community and sample) is selected for each group and summed
to provide the PMA value (Table 1).
4

-------
Table 1. Example calculation of a PMA score.
Order/Group
Model
Sample
Lesser Value
Trichoptera
17
8
8
Lepidoptera
2
0
0
Ephemeroptera
31
26
26
Coleoptera
18
30
18
Diptera
22
36
22
Odonata
4
0
0
Crustacea
4
0
0
Other
2
0
0
Total
100
100
74
Biological assessment categories for the PMA were selected using the percent range of deviation
from a reference condition, as proposed by EPA's National Rivers and Streams and Lakes
Assessments (USEPA 2006, USEPA 2010, USEPA 2013). Macroinvertebrate data collected
from reference sites were considered representative of a range of expected values for least-
disturbed by human activities and this distribution was used as the benchmark for setting
thresholds. The thresholds indicate distinct condition classes (e.g., good, fair, poor) and are
based on the degree of disturbance drawn from the reference condition (Figure 3).
Biological Condition (PMA Score)
100
go -
so -
40 -
Goat?
0	1	2
ReFerence Data
Figure 3. PMA score thresholds for biological condition categories (good, fair, poor).
5

-------
Results and Discussion
In general, for bioassessment measures to be useful, they must be able to make assessments of
water quality with some degree of accuracy, minimum variability, and reasonable level of
taxonomic effort (Barbour et al. 1992, Resh and Jackson 1993, Resh 1994, Hannaford and Resh
1995). In Puerto Rico, these issues are further compounded by the limited development of
bioassessment protocols, lack of information on invertebrate taxonomy and functional feeding
group classification, and limited knowledge about responses of aquatic invertebrates to water
pollution. To test the accuracy of the PMA, this measure was evaluated against a land use
gradient consisting of stream sites of reference quality (Stoddard et al. 2006) and those with high
deforestation. In comparing the land use response of the Mil and PMA, the PMA did not
distinguish reference from deforested sites with the same degree of separation (Figure 4). In
another test of accuracy, the PMA was tested for its response to water pollution in a small
urbanized watershed named the Rio Piedras. The Rio Piedras is known for water pollution
typical of highly urban watersheds, including chemical contamination from stormwater runoff
and raw sewage (Lugo et al. 2011, Potter et al. 2013, Ramirez et al. 2014). Both the PMA and
Mil tracked the changes of water quality going from the less developed headwaters to the highly
urban downstream reach (Figure 5). These findings are consistent with other studies of
biological and chemical conditions in the Rio Piedras (de Jesus-Crespo and Ramirez 2011).
Unlike the PMA developed for use in New York State (Novak and Bode 1992), the modified
version for Puerto Rico could not be tested for its correlation with indexes commonly used in
temperate North America, such as the Hillsenhoff biotic index and EPT index. First, the
response of these indexes to water pollution is unknown for Puerto Rico macroinvertebrates.
Second, pollution tolerance values have not been developed to use Puerto Rico taxa in a HBI.
Lastly, plecopterans are not found in Puerto Rico, thus limiting the use of the typically pollution
sensitive EPT index.
6

-------
PR PMA Distribution
PR Mil Distribution
100
I
Forest Cover
1-	>85% Reference Data (2006, 3-2009, 2011) (n = 42)
2-	<15% 1994 Sites (n = 26)
I
Forest Cover
1-	>85% Reference Data (2006, 3-2009, 2011) (n = 42)
2-<15%	1994 Sites (n = 26)
Figure 4: Box and whisker plots of the PMA and Mil for two different levels of forest cover
Mil and PMA for the Rio Piedras Sites
60
20
8 9 10
Rio Piedras Sites
Figure 5. Mil and PMA scores for the Rio Piedras sites.
7

-------
Sampling and seasonal variability of the index was examined. The sampling coefficient of
variation (CV) for 42 reference sites was determined to be 17%. This was greater than the 5.3%
for a set of twenty pristine sites, but similar to the CV of 16% using a raw data set of 50 sites
(Novak and Bode 1992) and generally lower than commonly used macroinvertebrate community
measures, such as, taxa richness (genus and species), Hillsenhoff biotic index, percent EPT, and
EPT richness (Barbour et al. 1992) and (Szczytko 1989).
Few examples of studies from the neotropics exist in the documented literature on the effects
season change has on the composition of benthic macroinvertebrates. Unlike temperate North
America where stream temperature and inputs of organic matter (e.g., leaves) respond abruptly to
seasonal change, in Puerto Rico stream temperatures and inputs of organic matter are relatively
uniform throughout the year. Bioassessment programs in temperate North America target a
specific index period to avoid the intra annual variability known to occur with seasonal
recruitment cycles of invertebrates. In Puerto Rico the climate is humid subtropical, with rainfall
weakly related to season and amounts generally greater with tropical disturbances September
through November. Seasonal variability was examined for six reference stream sites that were
sampled three times (spring, summer, and fall) for one year. Individual sites displayed
fluxuation in the index scores between seasons (Figure 6). Sites 6, 1, and 17 had more consistent
scores between seasons, while sites 47 and 1152 had sharp drops in PMA values during the
summer. Overall, PMA scores for site 7 were lower regardless of season, but increased slightly
during the summer. The mean coefficient of variation calculated for the PMA was 18% and this
compared favorably to the value of 14% obtained in New York State for a similar study of
seasonal variability (Novak and Bode 1992). Nonetheless, it is not recommended here that
biological conditions be compared from samples collected at different times of the year.
Seasonality Study PMA Scores

100.00

90.00

80.00

70.00
k.
o
60.00
o

(~)
50.00
<

2
40.00
Q-


30.00

20.00

10.00

0.00
¦47
¦17
7
¦6
¦1152
•1
Spring
Summer
Fall
Figure 6. Change in PMA scores at six Puerto Rico stream sites during the study period.
8

-------
The utility of the PMA index for application in Puerto Rico is the ease in which biological
conditions could be determined without the need for extensive information on invertebrate
taxonomy. Currently there is a lack of broad taxonomic keys and skills required by water quality
professionals to identify macroinvertebrates below the family-level of taxonomy. Some taxa are
well documented and others are less understood (Gutierrez-Fonseca et al 2013). Use of the PMA
index is not meant to discourage or substitute the development of low-level invertebrate
taxonomy in Puerto Rico, but serve has an interim measure until the science catches up to the
need. Genera and species-level taxonomy is necessary for the calculation of other community
indices. Another advantage of the index is the computation of the index score is relatively easy.
Most multimetric indices require more extensive data management programs to calculate index
scores.
With some modification of the PMA, this measure has potential for use throughout the
Caribbean and other regions of the Neotropical zone. The original community composition
model on which the PMA index was developed for streams in New York, was modified to take
into account zoogeographical differences of aquatic fauna found in Puerto Rico. Likewise, if this
model was developed for use in other islands of the Caribbean it would need to reflect any
geographical or sampling method differences.
Literature Cited
Barbour, M. T., J. L. Plafkin, B. P. Bradley, C. G. Graves, and R. W. Wisseman. 1992.
Evaluation of EPA's rapid bioassessment benthic metrics: metric redundancy and variability
among reference stream sites. Environmental Toxicology and Chemistry 11:437-449.
Bode, R.W., M. A. Novak, and L.E. Able. 1991. Methods for rapid bioassessment of streams.
New York State Department of Environmental Conservation, Division of Water, Albany, NY.
de Jesus-Crespo, R. and A. Ramirez. 2011. Effects of urbanization on stream physiochemistry
and macroinvertebrate assemblages in a tropical urban watershed in Puerto Rico. Journal of the
North American Benthological Society 30:739-750.
Gutierrez-Fonesca, P.E, K.G. Rosas, and A. Ramirez. 2013. Aquatic Insects of Puerto Rico: a list
of families. Dugesiana. 20(2):215-219.
Hannaford, M. J. and V. H. Resh. 1995. Variability in macroinvertebrate rapid-bioassessment
surveys and habitat assessments in a northern California stream. Journal of the North American
Benthological Society 14:430-439.
Lugo, A. E., Gonzales, O. M. R. and C. R. Pedraza. 2011. The Rio Piedras watershed and its
surrounding environment. Vol FS-980. US Department of Agriculture, Rio Piedras.
Novak M. A. and R. W. Bode. 1992. Percent model affinity: a new measure of
macroinvertebrate community composition. Journal of the North American Benthological
Society 11:80-85.
Olsen, A.R. and D.V. Peck. 2008. Survey design and extent estimates for the Wadeable Streams
9

-------
Assessment. Journal of the North American Benthological Society 27:822-836.
Potter, J. D., W. H. McDowell, A. M. Helton, and M. L. Daley. 2013. Incorporating urban
infrastructure into biochemical assessment of urban tropical streams in Puerto Rico.
Biochemistry.
Ramirez, A., K. G. Rosas, A. E. Lugo, and O. M. Ramos-Gonzalez. 2014. Spatio-temporal
variation in stream water chemistry in a tropical urban watershed. Ecology and Society 19:45.
Resh, V. H. and J. K. Jackson. 1993. Rapid assessment approaches to biomonitoring using
benthic macroinvertebrates. Pages 195-233 in D. M. Rosenberg and V. H. Resh (editors).
Freshwater biomonitoring and benthic macroinvertebrates. Chapman and Hall, New York.
Resh, V. H. 1994. Variability, accuracy, and taxonomic costs of rapid bioassessment
approaches in benthic macroinvertebrate biomonitoring. Bollettino di Zoologia 61:375-383.
Stoddard, J. L., D. P. Larsen, C. P. Hawkins, R. K. Johnson, and R. H. Norris. 2006. Setting
expectations for the ecological condition of streams: the concept of reference condition.
Ecological Applications 16:1267-1276.
Szczytko. S. W. 1989. Variability of commonly used macroinvertebrate community metrics for
assessing biomonitoring data and water quality in Wisconsin streams. Pages 12-22 in W. S.
Davis and T. P. Simon (editors). Proceedings of the 1989 Midwest Pollution Control Biologists
Meeting. EPA-905/9-89-007. U. S. Environmental Protection Agency, Region 5,
Environmental Services Division, Chicago, Illinois.
U.S. Environmental Protection Agency. 2006. Wadeable streams assessment: a collaborative
survey of the nation's streams. EPA 841/B-06/002. U.S. Environmental Protection Agency,
Washington, DC.
U.S. Environmental Protection Agency. 2010. National lakes assessment: a collaborative survey
of the nation's lakes. EPA 841-R-09-001. U.S. Environmental Protection Agency, Washington,
DC.
U.S. Environmental Protection Agency. 2013. National rivers and streams assessment 2008-
2009: a collaborative survey. EPA/841/D-13/001. U.S. Environmental Protection Agency,
Washington, DC.
10

-------