1
Supporting Information
2
Additional Supporting Information may be found in the online version of this article:
3
Appendix S1.
4
Additional Methodological Details and Results
5
Appendix S2.
6
Table S2.1 Summary of geographic distances between sites and variables used to create land-
7
use/cover attributes of dispersal pathways.
8
Table S2.2 Summary of environmental variables used to calculate environmental dissimilarity.
1
Appendix S1.
2
Additional Methodological Details and Results
3
Details on taxonomic changes
4
All specimens identified to non-insect taxa and those insect taxa identified to Collembola
5
(n = 5), Isotomidae (Collembola) (n = 2), Isotomurus (Collembola: Isotomidae) (n = 5),
6
Lepidoptera (n = 1), and Curculionidae (Coleoptera) (n = 1) were removed from the dataset prior
7
to analysis. All Collembola were removed due to their lack of a flight capable stage. All
8
Lepidoptera identified to order were removed because many fully aquatic lepidopteran larvae in
9
the study region are often identifiable to at least family by the presence of conspicuous gill
10
structures along their abdomen. Those without gill structures are generally terrestrial, and as a
11
result, those identified to order in the dataset were considered accidentally collected terrestrial
12
taxa (individuals identified to family and genus were retained). Aquatic Curculionidae do not
13
differ significantly from terrestrial Curculionidae, and all Curculionidae taxa were removed
14
because specimens included in the MBSS dataset may be terrestrial taxa collected accidentally.
15
The specimens from non-insect taxa (n = 1228) represented only 5.6% of the original dataset,
16
and the specimens from the flight incapable and potentially terrestrial taxa (Collembola,
17
Isotomidae, Isotomurus, Lepidoptera, and Curculionidae, n = 14) represented only 0.1% of the
18
original dataset.
19
As stated in the methods, some specimens were either removed or reassigned to a family
20
level designation to achieve taxonomic congruency between samples for measures of
21
dissimilarity between communities. The specimens identified to genus were changed to family if
22
the statistic:
23
x = 10 * (no. family / total abundance)
Eq. A1
24
where no. family = the abundance of all specimens identified to family from all samples and total
25
abundance = the total abundance of all specimens from the family regardless of taxonomic
26
designation, was greater than the number of genera recorded for the family being evaluated.
27
Specimens with family level identifications were removed from the data set if x was lower than
28
the number of genera recorded for the family being evaluated. For example, if 50 specimens
29
were identified to family out of 100 total specimens from that family (x = 10 * 50 / 100 = 5) and
30
only 3 genera were identified, then all specimens identified to genus were reclassified at the
31
family level (x >3). If 20 specimens were identified to family out of 100 total specimens from
32
that family (x = 10 * 20 / 100 = 2) and 5 genera were identified from that family, then the
33
specimens identified to family were deleted from the data set (x <5) and the genus level
34
identifications were retained. This statistic was created specifically for this study to minimize the
35
amount of community data lost, and abundances were examined for each family individually to
36
ensure that minimal data was lost. Thus, the rule was broken for one family where four
37
individuals (identified to family) would have been deleted if not for one specimen identified to
38
genus. In this case, all genera were reassigned to the family level when our rule would have
39
dictated that all specimens identified to family be deleted.
40
Of the 63 families of insects found in this study, 31 had specimens that either had their
41
taxonomic designations changed or specimens deleted from the dataset. Of these 31 families, 14
42
had specimens identified to genus reassigned to family. Eight of these 14 families had specimens
43
simultaneously identified to the family level and a single genus (n = 156, 0.8% of all specimens
44
retained for analysis), which meant little taxonomic resolution was lost when coalescing them all
45
to the family level. Six of these 14 families had specimens simultaneously identified to family
46
and multiple genera (n = 320, 1.6% of all specimens retained for analysis). An additional 17
47
families with specimens simultaneously identified to genus and family had the family level
48
specimens deleted from the original dataset (n = 635, 2.9% of all specimens in the original
49
dataset). Specimens from six Chironomidae tribes or subfamilies (n = 391, 1.8% of all specimens
50
in the original dataset) and four specimens identified to order were deleted from the original
51
dataset as well. The number specimens deleted from individual families ranged from 1-420
52
specimens (Chironomidae had the most individuals deleted) and comprised between 1.5%-38.6%
53
of the specimens belonging to that family.
54
Details on collection methods for environmental variables
55
All environmental variables used were collected by the MD-DNR for the MBSS
56
program. All water chemistry variables used in our analysis originated from field measurements
57
done at the top of the sampling reach where stream invertebrate samples were collected. Mean
58
stream width, mean thalweg depth, and mean velocity were calculated from measures at 4 evenly
59
spaced transects along the 75m reach (0m, 25m, 50m, and 75m). Qualitative measures of
60
environmental conditions in the stream (stream habitat, epifaunal substratum, velocity depth
61
diversity, pool quality, and riffle quality) were done by visually ranking each variable on a scale
62
of 0-20 with 0 representing the most poor and 20 representing the most optimal quality for each
63
measure in the 75m reach (see Stranko et al., 2007 for specific descriptions). Percent
64
embeddedness of the stream bottom and shading by the riparian vegetation were also visually
65
estimated.
66
Details on the use of conductivity over nutrient data for calculating environmental dissimilarity
67
Nutrient concentrations (e.g. nitrogen and phosphorus) and conductivity are important
68
predictors of overall water quality and are often correlated to urban land use (Roy et al., 2003;
69
King et al., 2005). The causal links between elevated nutrient concentrations, individual insect
70
fitness, and community composition are difficult to discern given 1) the number of confounding
71
factors and interactions that can occur in urban stream environments and 2) the numerous
72
indirect pathways that altered water chemistry affect insect populations and communities (King
73
et al., 2005; Yuan, 2010; Liess et al., 2012). Even without evidence of a direct effect, elevated
74
conductance can represent a surrogate for a suite of potentially confounded and interacting
75
environmental characteristics that can change the composition of stream insect communities
76
along a gradient of stream quality (Roy et al., 2003; King et al., 2005). Specific to this study,
77
streams draining urban and forested watersheds in Maryland’s Piedmont were found by Craig
78
(2009) to be nitrogen saturated, which indicated nitrogen may not be a good predictor of insect
79
habitat quality at high nitrogen concentrations. Additionally, King et al. (2005) found that
80
conductance was directly related to land use/cover and macorinvertebrate assemblages for MBSS
81
data in Maryland, but nitrogen concentrations were not. Phosphorus mobilization can have high
82
temporal and spatial variability (Withers & Jarvie, 2008), and as a result, we did not believe that
83
phosphorus data generated from grab samples used for the MBSS data described the quality of
84
stream habitat as well as measures of conductivity. Thus, we chose to use conductivity over
85
nutrient data for our analysis of environmental dissimilarity.
86
Model diagnostics
87
Variance inflation factors (VIF) for models indicated that overdispersion was not a
88
concern for the thirteen models fitted to the assemblage dissimilarity data (VIF ~1; Kutner,
89
Nachtsheim & Neter, 2004). Correlations among environmental dissimilarity (i.e., the variable
90
representing in-stream processes) and the predictors representing dispersal were also not a
91
concern. Pearson’s correlation between environmental dissimilarity and overland geographic
92
distance, corridor geographic distance, overland pathway land-use/cover attributes and
93
corridor pathway land-use/cover attributes were 0.13, 0.11, -0.25 and -0.22 respectively.
94
References
95
Craig, L.S. (2009) Nitrogen saturation in streams and forests of the Maryland Piedmont. PhD
96
97
Thesis, University of Maryland, College Park, MD, USA.
King R.S., Baker M.E., Whigham D.F., Weller D.E., Jordan T.E., Kazyak P.F. & Hurd M.K.
98
(2005) Spatial considerations for linking watershed land cover to ecological indicators in
99
streams. Ecological Applications, 15, 137-153.
100
101
102
Kutner M.H., Nachtsheim C.J. & Neter J. (2004) Applied Linear Regression Models. Fourth
edition. McGraw-Hill/Irwin, New York, NY.
Liess A., Le Gros A., Wagenhoff A., Townsend C.R. & Matthaei C.D. (2012) Landuse intensity
103
in stream catchments affects the benthic food web: consequences for nutrient supply,
104
periphyton C:nutrient ratios, and invertebrate richness and abundance. Freshwater
105
Science, 31, 813-824.
106
Roy A.H., Rosemond A.D., Paul M.J., Leigh D.S. & Wallace J.B. (2003) Stream
107
macroinvertebrate response to catchment urbanisation (Georgia, U.S.A). Freshwater
108
Biology, 48, 329-346.
109
Stranko S., Boward D., Kilian J., Becker A., Ashton M., Schenk A., Gauza R., et al. (2007)
110
Maryland Biological Stream Survey, Round Three Field Sampling Manual. Maryland
111
Department of Natural Resources Publication # 12-2162007-190. Annapolis, MD.
112
113
114
115
Withers P.J.A. & Jarvie H.P. (2008) Delivery and cycling of phosphorus in rivers: A review.
Science of the Total Environment, 400, 379-395.
Yuan L.L. (2010) Estimating the effects of excess nutrients on stream invertebrates from
observational data. Ecological Applications, 20, 110-125.
116
Appendix S2.
117
Table S2.1 Summary statistics for each MDE6 drainage unit of 1) geographic distances between sample site pairs and 2) GIS variables
118
used to create the land-use/cover attributes for overland and corridor pathways. Listed are the range, mean (𝑥̅ ), and standard
119
deviation (SD) for each variable.
Overland dispersal pathway
Distance (km)
% Foresteda
% Imperviousb
% Com hd-res-1c
Road int. (no. km-1)d
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Bush
0.8-22.7
9.3(4.7)
1.0-51.3
23.0(10.0)
5.5-90.3
49.2(16.8)
0.0-41.5
9.8(8.5)
0.0-6.5
3.3(1.2)
Gunpowder
0.9-53.2
21.4(12.1)
0.1-89.4
32.7(12.3)
0.2-94.2
18.5(13.6)
0.0-48.8
2.0(4.2)
0.0-8.0
1.4(0.8)
Patapsco
0.9-54.4
18.8(9.9)
0.0-99.5
30.5(12.3)
0.0-93.5
31.1(19.0)
0.0-48.0
6.8(7.6)
0.0-11.2 2.0(1.5)
Patuxent
0.8-33.2
13.9(6.6)
6.7-59.3
27.7(9.8)
1.9-71.9
17.8(12.1)
0.0-14.6
1.0(2.3)
0.4-4.6
1.4(0.7)
Susquehanna
0.5-43.3
17.0(9.8)
1.6-75.8
30.4(10.3)
0.7-35.8
8.8(3.9)
0.0-11.7
0.5(1.1)
0.0-2.4
1.0(0.2)
MDE6-DU
Corridor dispersal pathway
Distance (km)
% Forestede
% Imperviousb
% Com hd-res-1c
Road int. (no. km-1)d
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Bush
2.3-64.8
26.8(14.9)
9.1-81.2
50.2(12.2)
4.4-66.0
20.6(9.1)
0.0-17.7
4.3(3.6)
0.0-1.6
0.7(0.2)
Gunpowder
1.3-127.1 55.3(30.0)
6.1-97.4
59.4(14.0)
0.4-57.0
7.4(3.8)
0.0-13.4
0.4(0.8)
0.0-2.4
0.5(0.2)
Patapsco
1.5-132.4 59.0(32.1)
7.8-100
57.7(17.1)
0.0-63.1
17.4(11.8)
0.0-24.8
2.1(2.6)
0.0-2.4
0.5(0.4)
Patuxent
5.6-140.0 69.9(42.9)
23.7-95.0 73.4(12.3)
2.0-33.7
9.6(5.2)
0.0-11.8
1.9(1.7)
0.1-1.2
0.5(0.2)
Susquehanna
0.8-107.2 41.3(23.2)
0.0-99.0
0.9-19.4
7.8(2.4)
0.0-2.6
0.1(0.2)
0.0-1.0
0.3(0.1)
MDE6-DU
52.8(10.7)
120
a
% Forested is the percent forested land use calculated for a 100m buffer around the dispersal pathway
121
b
% Impervious is the percent impervious surfaces calculated for a 100m buffer around the dispersal pathway
122
c
% Com hd-res-1 is the percent commercial and high density residential land use in a 100m buffer around the dispersal
123
pathway
124
d
Road int. is the number of road-dispersal pathway interactions per km
125
e
% Forested is the percent forested land use calculated for a 30m buffer around the dispersal pathway
126
Table S2.2 Summary of environmental variables used to calculate environmental dissimilarity. Listed are the range, mean (𝑥̅ ), and
127
standard deviation (SD) for each variable.
Chemistry
DOa (mg L-1)
Condb (μS cm-1)
pH
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Bush
6.2-10.3
8.1(1.0)
6.4-7.9
7.0(0.4)
164-628
323.1(149.3)
Gunpowder
4.8-9.7
8.1(1.0)
6.3-7.3
7.3(0.4)
90-1000
244.3(164.1)
Patapsco
3.3-17.9
8.3(1.7)
5.2-9.1
7.3(0.5)
100-1330
331.3(263.5)
Patuxent
3.7-9.1
7.5(1.2)
6.5-7.7
6.9(0.3)
80-540
200.6(113.4)
Susquehanna
4.7-11.6
8.6(1.1)
6.2-8.5
7.0(0.5)
89-851
189.1(111.3)
MDE6-DU
Stream habitat
In-stream habitat
MDE6-DU
Range
𝑥̅ (SD)
Episubstratec
Range
𝑥̅ (SD)
Vel-depthd
Range
𝑥̅ (SD)
Pool quality
Range
𝑥̅ (SD)
Riffle quality
Range
𝑥̅ (SD)
Bush
2-16
11.8(3.9)
1-16
11.4(3.7)
2-15
9.6(3.3)
6-16
11.2(3.2)
2-14
8.7(3.6)
Gunpowder
3-18
12.5(4.0)
3-19
12.4(4.5)
4-15
9.4(2.5)
3-17
9.9(3.4)
0-17
11.0(4.2)
Patapsco
1-18
12.9(3.6)
2-18
13.3(3.9)
1-16
9.2(3.1)
3-16
9.5(3.2)
0-18
11.1(3.8)
Patuxent
1-17
12.1(4.7)
1-17
12.0(4.4)
2-14
10.1(3.6)
1-18
10.3(4.3)
1-16
11.0(4.2)
Susquehanna
6-17
12.7(2.8)
4-18
13.2(3.4)
6-17
10.2(2.8)
4-16
10.1(3.0)
4-17
12.3(2.7)
Physical characteristics
% Embedded
Range
𝑥̅ (SD)
Bush
0-70
29.9(20.1)
Gunpowder
9-95
Patapsco
0-100
Patuxent
15-100 35.8(20.2)
MDE6-DU
%Shading
Depthf (cm)
Velocityg (m s-1)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
Range
𝑥̅ (SD)
15.0-95.5 81.6(19.0)
0.5-6.2
2.6(1.5)
5.3-48.3
19.2(11.7)
0.01-0.28
0.11(0.07)
37.5(19.4)
5.0-98.0
77.3(22.8)
0.4-7.5
2.3(1.2)
5.0-58.8
17.4(9.2)
0.01-0.63
0.17(0.14)
33.6(18.9)
12.0-99.0 83.1(16.3)
0.7-7.9
2.7(1.5)
3.5-44.5
15.7(8.3)
0-0.49
0.17(0.12)
60.0-95.0
0.3-4.6
2.3(1.1)
3.3-36.5
16.5(8.9)
0.01-0.25
0.10(0.08)
Range
𝑥̅ (SD)
Widthe (m)
86.2(9.2)
Susquehanna
5-100
26.9(19.2)
30.0-99.0 81.3(16.7)
128
a
DO = dissolved oxygen
129
b
Cond = conductivity
130
c
Epi substrate = epifaunal substratum
131
d
Vel-depth = velocity depth diversity
132
e
Width = mean stream width
133
f
Depth = mean depth in thalweg
134
g
Velocity = mean velocity
0.7-5.9
2.7(1.4)
5.0-37.5
19.8(8.4)
0.02-0.80
0.20(0.14)