(dissenting).
I respectfully dissent. While I agree with my colleagues in the majority that the policy exclusion for losses “caused by rain” is an important aspect of a proper analysis of this case, I disagree that our disposition turns on the perceived distinction between rain and rainwater.
It is important to understand precisely the loss for which Amish Connection seeks coverage. Amish Connection does not claim coverage for the cost of repairing or replacing the failed drain pipe. Cf. Berry v. Commercial Union Ins. Co., 87 F.3d 387, 388 (9th Cir.1996) (noting the damage for which coverage was sought was a damaged irrigation pipe); Arkwright-Boston Mfrs. Mut. Ins. Co. v. Wausau Paper Mills Co., 818 F.2d 591, 593-95 (7th Cir.1987) (affirming summary judgment granted to the insurer based on a corrosion exclusion when the only damage was the cost of repairing a corroded reactor used to extract papermaking chemicals). Rather, the loss was water damage to Amish Connection’s inventory, furnishings, equipment, and documents after the pipe burst.
Water damage is among the “Specified Causes of Loss” covered by the policy. The policy definition of “water damage” expressly includes “accidental discharge or leakage of water or steam as the direct result of the breaking or cracking of any *244part of a system or appliance containing water or steam.” Thus, the casualty Amish Connection suffered is a covered loss unless it falls within one of the narrowly construed limitations or exclusions. See Farm & City Ins. Co. v. Gilmore, 539 N.W.2d 154, 157 (Iowa 1995) (“Insurance policy exclusions are strictly construed against the insurer.”); see also Essex Ins. Co. v. Fieldhouse, Inc., 506 N.W.2d 772, 776 (Iowa 1993) (“[W]hen an exclusion acts to withdraw a promised coverage, it must be clearly and explicitly defined.”); Grinnell Mut. Reins. Co. v. Emp’rs Mut. Cas. Co., 494 N.W.2d 690, 693 (Iowa 1993) (noting exclusionary clauses “require a narrow or restrictive construction”); Kalell v. Mut. Fire & Auto. Ins. Co., 471 N.W.2d 865, 867 (Iowa 1991) (“[Insurance policies are adhesion contracts, and exclusions will be strictly construed against the insurer.”).
My colleagues in the majority conclude as a matter of law the policy provides no coverage for Amish Connection’s loss because it was caused by rainwater. Because I believe a genuine issue of material fact remains on the question whether the loss was in fact caused by rain or, instead, by some other covered occurrence (or a combination of covered and uncovered occurrences), summary judgment is inappropriate and the case should proceed to trial.
“Causation has always been a troublesome concept for lawyers.” Banks McDowell, Causation in Contracts and Insurance, 20 Conn. L.Rev. 569, 569 (1988) [hereinafter McDowell]. Particularly troublesome are instances of “dual or concurrent causation — where two or more occurrences lead to a loss and at least one is covered under the terms of a single insurance policy.” Id. at 570 (footnotes omitted). In my view, this case presents one such instance. Amish Connection suffered a loss when a pipe burst during a heavy rainstorm. State Farm asserts the loss was “caused by rain” because the pipe would not have failed without the volume of water the storm produced. Yet, the summary judgment record in this case leaves room for a reasonable fact finder to find the loss Amish Connection suffered would not have occurred as a consequence of the rain alone, but only because the pipe failed while the rainwater was passing through it. In other words, a fact question remains whether the loss might not have occurred but for the confluence of at least two causes. See Randall L. Smith & Fred A. Simpson, Causation in Insurance Law, 48 S. Tex. L.Rev. 305, 312 (2006) (“Perils A and B, independent of each other, join in causing the loss, both perils being necessary to cause the loss since neither peril alone is sufficient to do so.”).
“While the notion of causation is simple, the factual world to which it must be applied is not.” McDowell, 20 Conn. L.Rev. at 573. Complexity arises “in situations of concurrent causation because one cause of the loss may be covered by an insurance policy and another cause either not covered or specifically excluded from coverage.” Erik S. Knutsen, Confusion About Causation in Insurance: Solutions for Catastrophic Losses, 61 Ala. L.Rev. 957, 962 (2010) [hereinafter Knutsen]. This case presents a causation and coverage quandary: Water damage caused by accidental leakage is a “Specified Cause of Loss” under the policy, but losses “caused by rain” are excluded.
In concurrent causation scenarios, some jurisdictions utilize a “liberal approach” to coverage under which an entire loss is covered as long as it was caused in part by a covered occurrence. See id. at 976 & n. 56 (listing the “minority of states” that follow this rule). Under this approach, if two events combine to cause the loss — one covered and one excluded — coverage exists even though the excluded event was a *245cause. While this rule “is not perfect,” it “promulgates a fairer result in instances where two ... concurrent causes produce the loss and neither on its own is sufficient to cause some harm.” Id. at 984; see also Brendan R. Vaughan, Note, Watered Down: Are Insurance Companies Getting Hosed in the Wind vs. Water Controversy?, 2008 U. Ill. L.Rev. 777, 796 (2008) (acknowledging a benefit of the liberal approach is its “ease of use in situations where it is difficult to segregate independent causes to determine which one is ... dominant”).
Another approach is the “efficient proximate cause” test under which the fact finder evaluates competing concurrent but-for causes to determine which was the primary cause of the loss. See Knutsen, 61 Ala. L.Rev. at 974-75 (explaining this approach). This test does require a decision about which of two competing causes was dominant. See Tento Int’l, Inc. v. State Farm Fire & Cas. Co., 222 F.3d 660, 662 (9th Cir.2000) (“While the rain may have been the most immediate cause of [the] damages, the more important inquiry involves determining ... the efficient proximate cause of the damage.”). In this case, the “efficient proximate cause” test would require the fact finder to determine whether the magnitude of the rain event or the condition or capacity of the drainpipe was the primary cause of the pipe failure and Amish Connection’s loss.
Finally, the “conservative approach” is a third analytical option. Under this approach, “if one cause in a causal chain is excluded from insurance coverage, the entire loss must be excluded from coverage, even if other causes may be covered by the policy.” Knutsen, 61 Ala. L.Rev. at 972.
Although we have not consistently applied any of these tests in Iowa, we have followed principles from both the liberal approach and the efficient proximate cause approach. See Grinnell Mut. Reins. Co., 494 N.W.2d at 691, 694 (concluding an insurer’s exclusion of coverage for bodily injury “arising out of the ownership, ... operation, use, loading or unloading of any automobile” did not preclude coverage for an injury sustained by a student while exiting a school bus because the school’s alleged “negligent supervision leading to negligent loading of the bus was not vehicle-related and was, or could have been determined to be, a proximate cause”); Kalell, 471 N.W.2d at 868 (“We hold that, when two independent acts are alleged, one vehicle-related and one not vehicle-related, coverage is still provided under the homeowners policy unless the vehicle-related negligence [that was excluded from coverage] is the sole proximate cause of the injury.”); Jordan v. Iowa Mut. Tornado Ins. Co. of Des Moines, 151 Iowa 73, 84, 130 N.W. 177, 181 (1911) (using the phrase “efficient cause” and noting the proposition “[t]hat other ... causes may also have contributed to the loss does not, of itself, relieve the [insurer] from responsibility”); Bettis v. Wayne Cnty. Mut. Ins. Ass’n, 447 N.W.2d 569, 571 (Iowa Ct.App.1989) (“[W]e look not necessarily to the last act in the chain of events, but rather to the predominant cause which set in motion the chain of events causing the loss.” (Emphasis added.)).
My research reveals no instance in which we have applied the approach styled as the “conservative approach.” This should not come as a surprise given our adoption of principles protecting insureds’ interests in litigation against their insurers. See Grinnell Mut. Reins. Co., 494 N.W.2d at 693 (“Coverage clauses are construed differently than exclusionary clauses. When construing coverage clauses, the words ... are given a broad, general and comprehensive meaning. Exclusionary clauses, however, require a narrow or re*246strictive construction.” (Citation omitted.)); Kalell, 471 N.W.2d at 867 (“If an insurance policy is ambiguous, we construe it in the light most favorable to the insured; insurance policies are adhesion contracts, and exclusions will be strictly construed against the insurer.”); N. Star Mut. Ins. Co. v. Holty, 402 N.W.2d 452, 454 (Iowa 1987) (“Where the meaning of terms in an insurance policy is susceptible to two interpretations, the one favoring the insured is adopted.”).
In my view, this is a case involving concurrent causes of a loss. There is evidence in this record tending to prove the rain event was.not the singular cause of Amish Connection’s loss. The summary judgment record leaves room for the possibility that a fact finder could determine Amish Connection’s loss would not have occurred but for the pipe’s weakened condition and its failure. When faced with multiple-cause scenarios like this, some courts — including our own — have found coverage as a matter of law when one of the causal factors was covered by an insurance policy. Kinney v. Farmers’ Mut. Fire & Ins. Soc’y of Kiron, 159 Iowa 490, 501-03, 141 N.W. 706, 710-11 (1913) (affirming the district court’s decision to grant the insured’s motion for directed verdict in a coverage dispute stemming from a winter windstorm featuring blowing snow because the insurer did not “show that the loss or damage was due to the [excluded hazards of] snow, or hail, ... or some other cause” rather than the covered hazard of wind); see also, e.g., Tento Int’l, 222 F.3d at 663-64 (concluding when the covered hazard of third-party negligence interacted with the excluded hazard of rain, the loss was covered as a matter of law under a State Farm policy very similar to the one at issue here); Berry, 87 F.3d at 392-93 (finding coverage when the covered hazard of third-party negligence caused irrigation pipes to deteriorate from a chemical reaction to a fungicide, even though the hazard of deterioration was otherwise excluded); Allstate Ins. Co. v. Smith, 929 F.2d 447, 451 (9th Cir.1991) (“We find that although rain ‘operate[d] more immediately in producing the disaster,’ it was the contractor’s failure to cover the premises that ‘set in motion’ the chain of events leading to Smith’s losses. The roofer’s failure to cover the exposed premises, therefore, was the efficient proximate cause of Smith’s losses.... [The insurer] must reimburse Smith for his losses to the extent provided by the policy.” (quoting Premier Ins. Co. v. Welch, 140 Cal.App.3d 720, 189 Cal.Rptr. 657, 660 (1983))); Safeco Ins. Co. of Am. v. Guyton, 692 F.2d 551, 555 (9th Cir.1982) (finding coverage for flood damage despite a flood exclusion in the policy because, had the negligently maintained flood control structures been properly maintained, they might not have collapsed when the flood struck them); Davidson Hotel Co. v. St. Paul Fire & Marine Ins. Co., 136 F.Supp.2d 901, 910 (W.D.Tenn.2001) (“[T]he covered peril of water, despite other factors leading to the loss which might have been excluded, renders St. Paul liable on this policy_ [T]he Court finds that summary judgment on the issue of coverage is warranted and GRANTS [the insured’s] motion for partial summary judgment on the issue of coverage.”); Mission Nat’l Ins. Co. v. Coachella Valley Water Dist., 210 Cal.App.3d 484, 258 Cal.Rptr. 639, 647 (1989) (reversing a jury verdict and judgment in the insurer’s favor, remanding with directions to enter judgment for the insured, and finding “coverage' because [the covered risk of] flooding contributed to or aggravated the loss, even though the excluded design defect was found to be the efficient proximate cause of the loss”).
In this case, however, we should not determine either coverage or exclusion as *247a matter of law because the causation question is not well-suited for summary judgment. A major insurance law treatise succinctly describes the strong preference for a jury to resolve causation questions:
Issues of causation are almost always questions of fact. Determining what perils were working and whether they were working together independently or in succession (and if so how), as well as identifying the type of damage following the operation of each peril, all fall to the fact finder.
5 New Appleman on Insurance Law § 44.06[1], at 44-45 (Jeffrey E. Thomas & Susan Randall eds., Library ed.2014). A case decided by the Washington Court of Appeals presents an apt illustration of this maxim. Sunbreaker Condo. Ass’n v. Travelers Ins. Co., 79 Wash.App. 368, 901 P.2d 1079, 1085 (1995). In the case, four possible causes of a loss were indicated. See id. at 1081. One of the potential causes of the loss was covered and three were excluded. See id. (noting damage from wind-driven rain was covered, but damage from dry rot, seepage, or defective construction was not). The court declined to resolve the coverage question as a matter of law because causation “is an issue of material fact precluding summary judgment.” Id. at 1085. The same principle should be applied here.
Caselaw from other jurisdictions supports my view that summary judgment should not have been granted in this case. See Mahaska Pork, L.P. v. Travelers Indem. Co. of Am., 777 F.Supp.2d 1185, 1192 (S.D.Iowa 2011) (“The question of ... proximate cause of the collapse is one that must ultimately be determined by a jury, not by this Court on motion for summary judgment.”); see also, e.g., Fidelity Coop. Bank v. Nova. Cas. Co., 726 F.3d 31, 38 (1st Cir.2013) (“The failure of the drain must properly be deemed the ‘efficient proximate cause’ of the damage, not the rain. The blocked or inadequate roof drain was a covered loss under the Policy ... and was not excluded by any other exclusion. Accordingly, it was error for the district court to conclude [as a matter of law] that the interior damage was ‘caused by rain’ and was excluded from coverage under the rain limitation provision.”); Paulucci v. Liberty Mut. Fire Ins. Co., 190 F.Supp.2d 1312, 1325 (M.D.Fla.2002) (refusing to grant summary judgment to the insurer because there remained a question of fact whether the loss was caused by the excluded hazard of preexisting rot or the covered hazard of tropical storm); Pioneer Chlor Alkali Co. v. Nat’l Union Fire Ins. Co. of Pittsburgh, 863 F.Supp. 1226, 1231-32 (D.Nev.1994) (refusing to grant summary judgment to the insurer because a jury could find either a covered event or an excluded event caused the loss); Garvey v. State Farm Fire & Cas. Co., 48 Cal.3d 395, 257 Cal.Rptr. 292, 770 P.2d 704, 714 (1989) (concluding “the question of causation is for the jury to decide” because the record contained sufficient evidence to support either of two causal possibilities); Simonetti v. Selective Ins. Co., 372 N.J.Super. 421, 859 A.2d 694, 700 (App.Div.2004) (“[T]here remains a factual question as to the actual cause of the damage to plaintiffs’ home.... Clearly, issues of causation are for the jury to resolve.”); Wider v. Heritage Maint., Inc., 14 Misc.3d 963, 827 N.Y.S.2d 837, 845 (Sup. Ct. 2007) (refusing to grant summary judgment in the insurer’s favor despite a rain limitation in the policy because the loss arguably caused by rain was also caused by a third party’s negligent" tarp placement); Villella v. Pub. Emps. Mut. Ins. Co., 106 Wash.2d 806, 725 P.2d 957, 964 (1986) (“[F]actual questions remain as to whether an alleged negligently constructed drainage system (a covered peril) was the efficient proximate cause of *248the loss. If so, the earth movement exclusionary clause would not exclude coverage.”); Lawver v. Boling, 71 Wis.2d 408, 238 N.W.2d 514, 521-22 (1976) (“Here i[t] has not yet been determined whether Lawyer's injuries resulted from negligence, if any, in the actual operation of the truck (an excluded risk) or from negligence in the choice of materials for and manner of construction of the rigging (a covered risk), or both.... That determination presents a question of fact which cannot be answered on a motion for summary judgment.”).
Some court decisions holding a coverage question could not be resolved as a matter of law are not procedurally identical to this one, but they are nonetheless supportive of my view that summary judgment was inappropriate here. For example, we have concluded at least twice that the causation question in insurance coverage disputes was properly presented to a fact finder. Crozier v. Lenox Mut. Ins. Ass’n, 252 Iowa 1176, 1184, 110 N.W.2d 403, 407-08 (1961) (“Plaintiff was insured against loss by windstorm. He claimed his loss was by windstorm. [The insurer] claimed there was no windstorm and that the barn fell because of faulty support during reconstruction. Fact issues for the jury were presented.”); Jordan, 151 Iowa at 84, 130 N.W. at 181 (affirming a trial court’s judgment after a bench trial in a dispute over whether the loss was caused by the covered hazard of windstorm or the excluded hazard of blowing snow, and noting causation “is a question of fact to be determined from the testimony”).
Courts in other jurisdictions have reached similar conclusions. See, e.g., Shinrone, Inc. v. Ins. Co. of N. Am., 570 F.2d 715, 716, 719 (8th Cir.1978) (affirming a jury verdict and judgment in a dispute over whether a herd of cattle perished from the covered hazard of wind or the excluded hazard of extreme temperature, and noting the causation question was “properly submitted to the jury for the determination of whether the windstorm was the dominant or efficient cause of the loss, even though there may have been other contributing causes”); Lynch v. Travelers Indem. Co., 452 F.2d 1065, 1066-67 (8th Cir.1972) (affirming a jury verdict and judgment in a dispute over whether a skating rink collapsed due to the covered hazard of wind or the excluded hazard of accumulated snow on the roof, because “[i]n the context of th[e] case, ... the trial court provided a proper framework for the jury to determine” which hazard caused the loss); State Farm Fire & Cas. Co. v. Von Der Lieth, 54 Cal.3d 1123, 2 Cal. Rptr.2d 183, 820 P.2d 285, 292 (1991) (reinstating a jury’s determination on causation and concluding it was error for an intermediate appellate court to determine the question as a matter of law); Davis v. United Servs. Auto. Ass’n, 223 Cal.App.3d 1322, 273 Cal.Rptr. 224, 230 (1990) (“Under the theory of the Davises’ expert, the cause of the loss was the failure to adequately prepare the soil and foundation: but for the negligence in preparing the soils and foundation, the earth movement would not have caused a loss. Under the theory of USAA’s expert, but for the earth movement, there would not have been a loss. In other words, there was a conflict in the evidence and the trial court was entitled to resolve this conflict.... ”); Gillis v. Sun Ins. Office, Ltd., 238 Cal.App.2d 408, 47 Cal.Rptr. 868, 875 (1965) (affirming the judgment entered after a fact finder determined wind (a covered hazard) caused the loss despite contribution from waves (an excluded hazard), and noting “[i]t may be inferred that in the absence of the blow from the gangway [blown down by wind], the docking facility would have ridden out the storm with no appreciable damage”); Wallach v. Rosenberg, 527 *249So.2d 1386, 1387-88 (Fla.Dist.Ct.App.1988) (insisting “the question as to which of several causes contributing to a loss is the ... proximate cause[ ] is one for the jury” and concluding “[wjhere weather perils combine with human negligence to cause a loss, it seems logical and reasonable” to allow the jury to find coverage “even if one of the causes is excluded from coverage”); Vormelker v. Oleksinski, 40 Mich.App. 618, 199 N.W.2d 287, 292-93 (1972) (“[Tjhere is a question of what the proximate cause of the damage was: the earth movement or the inadequate foundation. Under these circumstances, the trial court was correct in refusing to grant either a directed verdict or a judgment n.o.v.”). Together, these cases — both those that involved summary judgment and those in which a fact finder decided causation — demonstrate that concurrent causation issues often present factual questions only a fact finder should answer.
I acknowledge that some courts have granted summary judgment to the insurer in concurrent causation circumstances. When granted in those cases, summary judgment was typically founded on anti-concurrent-cause language in the particular insurance policy. See, e.g., TNT Speed & Sport Ctr., Inc. v. Am. States Ins. Co., 114 F.3d 731, 733-34 (8th Cir.1997) (affirming summary judgment in the insurer’s favor because although the loss was caused both by the covered hazard of vandalism and the excluded hazard of flood, specific language in the policy provided for exclusion in concurrent causation situations); Front Row Theatre, Inc. v. Am. Mfr.’s Mut. Ins. Cos., 18 F.3d 1343, 1349 (6th Cir.1994) (affirming summary judgment in the insurer’s favor because of express policy language excluding coverage from concurrently caused loss even if one cause was otherwise covered); Freedman v. State Farm Ins. Co., 173 Cal.App.4th 957, 93 Cal.Rptr.3d 296, 301 (2009) (affirming summary judgment granted to State Farm because of express policy language that excluded coverage for third-party negligence if it interacted in any way with an otherwise-excluded cause). The effect, if any, of anticoncurrent-cause language was not raised or decided in the district court and is not before us in this appeal.
The majority characterizes the “LOSSES INSURED” section of Amish Connection’s policy as a functional equivalent of anticoncurrent-cause language. The LOSSES INSURED section of the policy states, in its entirety:
We insure for accidental direct physical loss to property covered under this policy unless the loss is:
1. limited in the PROPERTY SUBJECT TO LIMITATIONS section; or
2. excluded in the LOSSES NOT IN- • SURED section that follows.
In my view, however, this is not an anti-concurrent-cause provision. Cases cited by the majority in this case illustrate the fact that anticoncurrent-cause provisions typically include the word “concurrent” — for example: “We will not pay for loss or damage caused directly or indirectly by any of the following. Such loss or damage is excluded regardless of any other cause or event that contributes concurrently or in any sequence to the loss.” TNT Speed & Sport Ctr., 114 F.3d at 732 (quoting an anticoncurrent-cause policy provision); see also Am. Fam. Mut. Ins. Co. v. Corrigan, 697 N.W.2d 108, 114-15 (Iowa 2005) (enforcing an anticoncurrent-cause provision).
Amish Connection’s policy does include express anticoncurrent-cause provisions like those involved in TNT Speed and Corrigan. For example, paragraph 1 of the “LOSSES NOT INSURED” section reads:
1. We do not insure under any coverage for any loss which would not *250have occurred in the absence of one or more of the following excluded events. We do not insure for such loss regardless of: (a) the cause of the excluded event; or (b) other causes of the loss; or (c) whether other causes acted concurrently or in any sequence with the excluded event to produce the loss[.]
The “following excluded events” to which this provision refers are: enforcement of any ordinance or law; earth movement; volcanic eruption; water such as flood, surface water, waves, tides, mudslides, sewer backup, and seeping groundwater; seizure or destruction of property by order of governmental authority; nuclear hazard; failure of power to the premises; and war or military action.
Paragraph 3 of the LOSSES NOT INSURED section contains additional anti-concurrent-cause language:
We do not insure for any coverage for any loss consisting of one or more of the items below. Further, we do not insure for loss described in paragraphs 1. or 2. immediately above regardless of whether one or more of the following: (a) directly or indirectly cause, contribute to or aggravate the loss; or (b) occur before, at the same time, or after the loss or any other cause of the loss[.]
State Farm did not raise either of these provisions in its motion for summary judgment. Nor did it raise them in its resistance to Amish Connection’s motion for more specific ruling. Instead, State Farm relied exclusively on language from a different section: the LOSSES INSURED section. Notably, State Farm expressly argued the district court need not consider any language from the LOSSES NOT INSURED section. The district court agreed, and even when it denied Amish Connection’s motion for a more specific ruling, the court expressly relied only on the LOSSES INSURED section of the policy.
The majority concludes State Farm’s reliance on the LOSSES INSURED policy provision effectively raised any anticoncur-rent-cause provisions in the policy. I strongly disagree. Both anticoncurrent— cause provisions are in the LOSSES NOT INSURED section — and State Farm expressly asserted the district court should not even consider language from that section in making its summary judgment decision.
I acknowledge the maxim that “[t]he individual provisions of a policy must be read in conjunction with the policy as a whole.” Ferguson v. Allied Mut. Ins. Co., 512 N.W.2d 296, 300 (Iowa 1994); accord Grinnell Mut. Reinsurance Co., 494 N.W.2d at 692. But this maxim does not authorize State Farm to sing a completely different song on appeal than it sang in the district court. See State v. Rutledge, 600 N.W.2d 324, 325 (Iowa 1999). Further, even if we were to consider policy provisions — the LOSSES NOT INSURED section — not relied upon by State Farm in the district court or the court itself in its summary judgment ruling, we should not find them sufficient to decide the coverage question in this case as a matter of law.
Amish Connection is not covered if its loss was “caused by rain,” as stated in the PROPERTY SUBJECT TO LIMITATIONS section, or if it is excluded in the LOSSES NOT INSURED section. I acknowledge that the word “or” is disjunctive in this context. But whether the loss was “caused by rain” — or by some other covered cause — is the fighting issue in this case. In my view, provisions in the LOSSES NOT INSURED section leave wide open the possibility that causes of loss other than rain are not excluded — even when concurrent causes are involved — and are therefore still in play here. A jury *251should be entitled to resolve factual questions like these.
I begin with paragraph 1 of the LOSSES NOT INSURED section. Paragraph 1 excludes any loss concurrently caused, even in part, by occurrences from a list of “excluded events.” These events are predominantly large-scale disasters — earthquakes, volcanic eruptions, nuclear meltdowns. Paragraph 1(d) includes “surface water” as an excluded event. However, because surface water is included in a list alongside flood, waves, tides, mudslides, and groundwater seepage, it is evident— using the canon of nosdtur a sodis — that surface water in paragraph 1(d) does not include the rainfall in this case. See Black’s Law Dictionary 1224 (10th ed.2014) (explaining that nosdtur a sociis means the definition of a word or phrase “should be determined by the words immediately surrounding it”). None of the other “excluded events” in paragraph 1 are remotely applicable to this case. Accordingly, paragraph 1 does not preclude the possibility that, if Amish Connection’s loss had multiple causes — one excluded and one covered — a jury could find the covered hazard was predominant and should afford Amish Connection coverage in this case.
I now turn to paragraph 3. This paragraph’s language must be unpacked carefully. The first sentence states: “We do not insure for any coverage for any loss consisting of one or more of the items below.” (Emphasis added.). The “items below” are subparts 3(a) and 3(b). 3(a) refers to “conduct, acts or decisions, including the failure to act or decide, of any person ..., whether intentional, wrongful, negligent or without fault” — essentially third-party negligence. 3(b) refers to “faulty, inadequate, unsound or defective” planning, design, construction, materials, and maintenance. But, as I have explained, Amish Connection’s loss does not consist of third-party negligence or faulty maintenance; if it did, the only costs for which Amish Connection would be seeking coverage would be the cost of repairing or replacing the failed pipe. Therefore, this first sentence of paragraph 3, standing alone, would not preclude coverage for a loss “caused by” — not “consisting of’— faulty maintenance.
The next sentence of paragraph 3 provides:
[W]e do not insure for loss described in paragraphs 1. and 2. immediately above' regardless of whether [the causes described in 3(a) and 3(b) ]: (a) directly or indirectly cause, contribute to or aggravate the loss; or (b) occur before, at the same time, or after the loss or any other cause of the loss[.]
As I have explained, paragraph 1 immediately above this language contains “excluded events” that are all inapplicable to the factual circumstances of this case. Paragraph 2 “immediately above” describes sixteen types of loss, in subparts 2(a) through 2(p). Only two of these could possibly be relevant here. Paragraph 2(k) — in, conjunction with paragraph 3 — excludes coverage for loss caused by “continuous or repeated seepage or leakage of water that occurs over a period of time” when that seepage combines with third-party negligence or faulty maintenance. However, the loss (water-soaked and ruined inventory) claimed in this case by Amish Connection was not due to continued or repeated seepage or leakage; it was due to a onetime.event — the .burst pipe.
Paragraph 2(d) is the only other subpart of paragraph 2 that could possibly be relevant here. It excludes loss caused by— not merely consisting of — rust, corrosion, and deterioration. Combined with paragraph 3, paragraph 2(d) could, at first blush, reasonably be read to exclude coverage if faulty maintenance caused the pipe *252to rust, which in turn caused the water damage in this case. However, paragraph 2(d) contains an additional statement: “But if accidental direct physical loss by any of the ‘Specified Causes of Loss’ ... results, we will pay for that resulting Zoss[.]” (Emphasis added.). In other words, although most loss caused by rust, corrosion, or deterioration is excluded, “Specified Causes of Loss” are still covered. As I have noted above, the Specified Causes of Loss under this policy include “water damage” arising from “accidental discharge or leakage of water or steam as the direct result of the breaking or cracking of any part of a system or appliance containing water or steam.” Thus, even if there is evidence that faulty maintenance of the pipe allowed it to rust and ultimately fail, Amish Connection’s loss is not excluded from coverage under paragraph 3(b).
My colleagues in the majority stress the fact that, according to State Farm, the policy was never intended to cover faulty maintenance. Whatever State Farm’s intent, -however, the actual policy language belies that assumption. The final clause of paragraph 8 states: “But if accidental direct physical loss results from items 3.a. and 3.b., we will pay for that resulting loss unless the resulting loss is itself one of the losses not insured in this section.” Item 3(b) is faulty maintenance. Thus, this express policy language indicates State Farm will pay for loss resulting from faulty maintenance unless the resulting loss is not insured in the LOSSES NOT INSURED section. Crucially, the rain limitation is not found in the LOSSES NOT INSURED section. In sum, even if State Farm and the district court had relied on the LOSSES NOT INSURED section of the policy — and here they clearly did not— summary judgment would not have been appropriate in this case.
Although the majority attempts to distinguish it, the Ninth Circuit’s decision in Tentó International is actually quite analogous to this case. See Tento Int’l, 222 F.3d at 661. A roofing contractor was making repairs but failed to install a temporary covering. Id. “Almost predictably, rain fell and damaged Tento’s electronics equipment.” Id. Tento’s insurance policy with State Farm contained a rain limitation identical to the one in Amish Connection’s policy, and it also contained an identical paragraph 3. See id. at 661-62. The court determined the efficient cause, not merely the most immediate cause, should determine coverage — even though the policy had several provisions materially identical to the ones in Amish Connection’s policy. See id. at 662.
Further, the court examined the “inexact wording in the resulting-loss provision.” Id. It determined Tentó had coverage as a matter of law:
The resulting-loss provision states that, “if accidental direct physical loss results from items 3.a. and 3.b. [i.e., the contractor’s third-party negligence], we will pay for that resulting loss unless the resulting loss is itself one of the losses not insured in this section.” (emphasis added). The question becomes whether the resulting loss — damage to Tento’s goods by rain — is “one of the losses not insured in this section.” (emphasis added). The words “in this section” appear in numbered paragraph 3 of the LOSSES NOT INSURED section of the policy. Logically, then, the words “in this section” refer to the LOSSES NOT INSURED section and that section does not preclude coverage for rain damage.
We recognize it is arguable that the scope of the “in this section” phrase could refer to the entire basic coverage of Section I, which includes the LOSSES INSURED as well as the LOSSES NOT *253INSURED sections. The LOSSES INSURED section incorporates by reference the earlier PROPERTY SUBJECT TO LIMITATIONS section, and that section includes the rain-damage limitation. Under this reading, the policy would not provide coverage for Tento’s rain-damaged goods. We reject this reading, however, because it is illogical, and defies a common-sense reading of the policy. Moreover, we note that when the policy refers to the entire “Section I,” it uses an upper case “S” to signal this broader reference.
Id. at 663. Amish Connection’s policy also uses an uppercase S when referring to the entire Section I. Applying the Tentó analysis to the facts of this case, Amish Connection might have coverage as a matter of law — although its loss might fit more squarely under the paragraph 8(b) faulty maintenance provision than under the paragraph 3(a) third-party negligence provision.
But I do not advocate we go that far on this record. Rather, I conclude the summary judgment should be reversed and the case remanded for trial. The evidence will establish whether Amish Connection’s loss was due to the mail’s faulty maintenance, or rust, decay, or some other cause or combination of causes. Although coverage is excluded if the loss was caused by rain, it could also have been caused by something else, or by a combination of other causes for which coverage exists.
When reviewing summary judgment rulings, we must make all factual inferences supported by the record in the nonmov-ant’s favor. See, e.g., Hagenow v. Am. Fam. Mut Ins. Co., 846 N.W.2d 373, 376 (Iowa 2014); Pitts v. Farm Bureau Life Ins. Co., 818 N.W.2d 91, 96 (Iowa 2012); Emp’rs Mut. Cas. Co. v. Van Haaften, 815 N.W.2d 17, 22 (Iowa 2012). In this case, I believe a reasonable fact finder could find rain was neither the sole nor even the primary factual cause of Amish Connection’s loss. I would therefore affirm the court of appeals and afford Amish Connection the opportunity to make its case on the causation question.
Wiggins and appel, jj., join this dissent.